diff --git "a/checkpoint-2000/trainer_state.json" "b/checkpoint-2000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2000/trainer_state.json"
@@ -0,0 +1,12019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.513248219670238,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.9189,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.8871,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 2.09,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.0244,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.9614,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 2.0378,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 2.0336,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.8758,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 1.9557,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.8357,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.8175,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 1.8704,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.8078,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.7974,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.7014,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.7933,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.899,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-05,
+      "loss": 1.8202,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-05,
+      "loss": 1.5872,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 1.6469,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4e-05,
+      "loss": 1.852,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.7968,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.8162,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6e-05,
+      "loss": 1.796,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001,
+      "loss": 1.7156,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 1.9844,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 1.5315,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 1.6643,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000116,
+      "loss": 1.8375,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012,
+      "loss": 2.052,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000124,
+      "loss": 1.7656,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 1.8665,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000132,
+      "loss": 1.7569,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 1.7503,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014,
+      "loss": 1.6777,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000144,
+      "loss": 1.7534,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000148,
+      "loss": 1.8081,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000152,
+      "loss": 1.6412,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 1.7198,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016,
+      "loss": 1.6977,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000164,
+      "loss": 1.8538,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000168,
+      "loss": 1.8738,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000172,
+      "loss": 1.8707,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 1.7029,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018,
+      "loss": 1.6269,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 1.7867,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000188,
+      "loss": 1.6741,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000192,
+      "loss": 1.8431,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000196,
+      "loss": 1.7986,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.6245,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020400000000000003,
+      "loss": 1.8149,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020800000000000001,
+      "loss": 1.831,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021200000000000003,
+      "loss": 1.6757,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021600000000000002,
+      "loss": 1.6475,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022000000000000003,
+      "loss": 1.7828,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022400000000000002,
+      "loss": 1.8008,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 1.6928,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000232,
+      "loss": 1.6002,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000236,
+      "loss": 1.7098,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00024,
+      "loss": 1.6816,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000244,
+      "loss": 1.6427,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000248,
+      "loss": 1.9189,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000252,
+      "loss": 1.7065,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00025600000000000004,
+      "loss": 1.7845,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 1.7162,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000264,
+      "loss": 1.8621,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000268,
+      "loss": 1.7064,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 1.6587,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000276,
+      "loss": 1.6538,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00028,
+      "loss": 1.788,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000284,
+      "loss": 1.8786,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000288,
+      "loss": 1.6963,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000292,
+      "loss": 1.7997,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000296,
+      "loss": 1.6258,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00030000000000000003,
+      "loss": 1.5675,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000304,
+      "loss": 1.8501,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000308,
+      "loss": 1.7404,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00031200000000000005,
+      "loss": 1.6924,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00031600000000000004,
+      "loss": 1.4889,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00032,
+      "loss": 1.6161,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000324,
+      "loss": 1.7842,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000328,
+      "loss": 1.7691,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000332,
+      "loss": 1.6911,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000336,
+      "loss": 1.6236,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00034,
+      "loss": 1.5516,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000344,
+      "loss": 1.702,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000348,
+      "loss": 1.7039,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00035200000000000005,
+      "loss": 1.7233,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00035600000000000003,
+      "loss": 1.7784,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00036,
+      "loss": 1.7119,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000364,
+      "loss": 1.7628,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00036800000000000005,
+      "loss": 1.7374,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00037200000000000004,
+      "loss": 1.5799,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000376,
+      "loss": 1.7434,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00038,
+      "loss": 1.8045,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000384,
+      "loss": 1.7939,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000388,
+      "loss": 1.7483,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000392,
+      "loss": 1.6368,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 1.6036,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0004,
+      "loss": 1.6768,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999999315068227,
+      "loss": 1.7205,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999997260273377,
+      "loss": 1.6885,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999993835616856,
+      "loss": 1.7387,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039999890411010117,
+      "loss": 1.7599,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999982876729127,
+      "loss": 1.7746,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039999753425054243,
+      "loss": 1.7643,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999966438435064,
+      "loss": 1.6161,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999956164524145,
+      "loss": 1.749,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999944520779703,
+      "loss": 1.7299,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999931507209715,
+      "loss": 1.6908,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039999171238230927,
+      "loss": 1.7822,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999901370629689,
+      "loss": 1.7854,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039998842476402934,
+      "loss": 1.644,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999865754866634,
+      "loss": 1.7075,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999845892321377,
+      "loss": 1.8361,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039998246600181267,
+      "loss": 1.6325,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999802057971426,
+      "loss": 1.5407,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039997780861967554,
+      "loss": 1.6327,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999752744710534,
+      "loss": 1.8547,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399972603353012,
+      "loss": 1.6868,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999697952673808,
+      "loss": 1.5997,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999668502160831,
+      "loss": 1.8115,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039996376820113615,
+      "loss": 1.7017,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999605492246508,
+      "loss": 1.716,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039995719328883204,
+      "loss": 1.7195,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039995370039597823,
+      "loss": 1.5336,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999500705484818,
+      "loss": 1.5242,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000399946303748829,
+      "loss": 1.7638,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039994239999959993,
+      "loss": 1.6565,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999383593034683,
+      "loss": 1.719,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039993418166320164,
+      "loss": 1.782,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039992986708166145,
+      "loss": 1.6487,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00039992541556180283,
+      "loss": 1.4585,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999208271066749,
+      "loss": 1.7476,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999161017194203,
+      "loss": 1.665,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003999112394032757,
+      "loss": 1.8235,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039990624016157135,
+      "loss": 1.7151,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003999011039977315,
+      "loss": 1.636,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998958309152741,
+      "loss": 1.7582,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998904209178107,
+      "loss": 1.8764,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039988487400904687,
+      "loss": 1.6772,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039987919019278184,
+      "loss": 1.7488,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998733694729087,
+      "loss": 1.6507,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039986741185341415,
+      "loss": 1.6778,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998613173383788,
+      "loss": 1.6078,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399855085931977,
+      "loss": 1.7432,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998487176384768,
+      "loss": 1.6079,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039984221246224005,
+      "loss": 1.5346,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998355704077224,
+      "loss": 1.4795,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039982879147947313,
+      "loss": 1.7697,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998218756821354,
+      "loss": 1.616,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039981482302044603,
+      "loss": 1.5246,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998076334992356,
+      "loss": 1.6524,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003998003071234285,
+      "loss": 1.5095,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997928438980427,
+      "loss": 1.5523,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997852438281901,
+      "loss": 1.905,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039977750691907614,
+      "loss": 1.6569,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997696331760002,
+      "loss": 1.6851,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039976162260435517,
+      "loss": 1.4957,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039975347520962776,
+      "loss": 1.6837,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039974519099739836,
+      "loss": 1.6918,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039973676997334104,
+      "loss": 1.7119,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997282121432238,
+      "loss": 1.6749,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399719517512908,
+      "loss": 1.9509,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003997106860883489,
+      "loss": 1.7915,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039970171787559553,
+      "loss": 1.554,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996926128807904,
+      "loss": 1.7785,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039968337111016984,
+      "loss": 1.6035,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996739925700638,
+      "loss": 1.6602,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399664477266896,
+      "loss": 1.6416,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996548252071837,
+      "loss": 1.7058,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000399645036397538,
+      "loss": 1.7482,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039963511084466354,
+      "loss": 1.6613,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00039962504855535854,
+      "loss": 1.553,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0003996148495365151,
+      "loss": 1.793,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039960451379511883,
+      "loss": 1.547,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039959404133824893,
+      "loss": 1.6133,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995834321730785,
+      "loss": 1.6491,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039957268630687386,
+      "loss": 1.6654,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039956180374699533,
+      "loss": 1.6804,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995507845008967,
+      "loss": 1.5704,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995396285761254,
+      "loss": 1.7928,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003995283359803225,
+      "loss": 1.8712,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039951690672122267,
+      "loss": 1.6735,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039950534080665414,
+      "loss": 1.6005,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039949363824453866,
+      "loss": 1.5399,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994817990428919,
+      "loss": 1.6491,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039946982320982274,
+      "loss": 1.6356,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039945771075353397,
+      "loss": 1.7414,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994454616823216,
+      "loss": 1.7381,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003994330760045756,
+      "loss": 1.5481,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039942055372877913,
+      "loss": 1.7216,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039940789486350925,
+      "loss": 1.7587,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039939509941743634,
+      "loss": 1.5533,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993821673993244,
+      "loss": 1.6301,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039936909881803097,
+      "loss": 1.8665,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039935589368250723,
+      "loss": 1.5792,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993425520017977,
+      "loss": 1.8631,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993290737850406,
+      "loss": 1.5977,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003993154590414675,
+      "loss": 1.638,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039930170778040364,
+      "loss": 1.772,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039928782001126767,
+      "loss": 1.8287,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039927379574357174,
+      "loss": 1.6486,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039925963498692156,
+      "loss": 1.8322,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992453377510163,
+      "loss": 1.5798,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992309040456485,
+      "loss": 1.8043,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039921633388070437,
+      "loss": 1.7151,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003992016272661634,
+      "loss": 1.7066,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039918678421209863,
+      "loss": 1.602,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003991718047286765,
+      "loss": 1.6753,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039915668882615715,
+      "loss": 1.7632,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00039914143651489367,
+      "loss": 1.7793,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000399126047805333,
+      "loss": 1.6199,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003991105227080153,
+      "loss": 1.658,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039909486123357427,
+      "loss": 1.6492,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003990790633927368,
+      "loss": 1.6837,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039906312919632357,
+      "loss": 1.6572,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039904705865524817,
+      "loss": 1.6667,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000399030851780518,
+      "loss": 1.6789,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003990145085832336,
+      "loss": 1.692,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003989980290745889,
+      "loss": 1.8569,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039898141326587136,
+      "loss": 1.6396,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003989646611684616,
+      "loss": 1.7412,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039894777279383356,
+      "loss": 1.7172,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039893074815355476,
+      "loss": 1.7643,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039891358725928594,
+      "loss": 1.6603,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988962901227811,
+      "loss": 1.7529,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988788567558875,
+      "loss": 1.6443,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000398861287170546,
+      "loss": 1.6134,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988435813787904,
+      "loss": 1.6467,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000398825739392748,
+      "loss": 1.6458,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003988077612246394,
+      "loss": 1.6152,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003987896468867784,
+      "loss": 1.6081,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039877139639157204,
+      "loss": 1.6888,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003987530097515207,
+      "loss": 1.6167,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000398734486979218,
+      "loss": 1.756,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003987158280873507,
+      "loss": 1.5726,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986970330886989,
+      "loss": 1.702,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986781019961359,
+      "loss": 1.5532,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003986590348226282,
+      "loss": 1.7834,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039863983158123557,
+      "loss": 1.7149,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039862049228511087,
+      "loss": 1.7413,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039860101694750015,
+      "loss": 1.7384,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039858140558174274,
+      "loss": 1.523,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039856165820127107,
+      "loss": 1.5346,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003985417748196108,
+      "loss": 1.7447,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003985217554503807,
+      "loss": 1.6395,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003985016001072925,
+      "loss": 1.7433,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003984813088041515,
+      "loss": 1.6971,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003984608815548556,
+      "loss": 1.6549,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039844031837339624,
+      "loss": 1.6979,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00039841961927385773,
+      "loss": 1.7413,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003983987842704176,
+      "loss": 1.6538,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039837781337734627,
+      "loss": 1.7508,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039835670660900757,
+      "loss": 1.669,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000398335463979858,
+      "loss": 1.4171,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039831408550444747,
+      "loss": 1.6787,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982925711974187,
+      "loss": 1.5834,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982709210735075,
+      "loss": 1.5367,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982491351475428,
+      "loss": 1.5698,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039822721343444635,
+      "loss": 1.5576,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003982051559492332,
+      "loss": 1.7985,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003981829627070111,
+      "loss": 1.8064,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003981606337229809,
+      "loss": 1.4714,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039813816901243645,
+      "loss": 1.6162,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003981155685907647,
+      "loss": 1.5121,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039809283247344517,
+      "loss": 1.8091,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003980699606760506,
+      "loss": 1.72,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003980469532142467,
+      "loss": 1.5372,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039802381010379194,
+      "loss": 1.7935,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039800053136053784,
+      "loss": 1.4957,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003979771170004287,
+      "loss": 1.4728,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003979535670395017,
+      "loss": 1.6232,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039792988149388704,
+      "loss": 1.7514,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003979060603798077,
+      "loss": 1.5872,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003978821037135795,
+      "loss": 1.4762,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003978580115116112,
+      "loss": 1.6576,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003978337837904041,
+      "loss": 1.8939,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003978094205665527,
+      "loss": 1.5578,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039778492185674414,
+      "loss": 1.6628,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003977602876777584,
+      "loss": 1.508,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003977355180464681,
+      "loss": 1.6372,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039771061297983883,
+      "loss": 1.5009,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003976855724949288,
+      "loss": 1.7411,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039766039660888905,
+      "loss": 1.7966,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003976350853389634,
+      "loss": 1.8018,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039760963870248835,
+      "loss": 1.6935,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039758405671689304,
+      "loss": 1.6938,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003975583393996994,
+      "loss": 1.7298,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00039753248676852206,
+      "loss": 1.6988,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0003975064988410682,
+      "loss": 1.6438,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000397480375635138,
+      "loss": 1.6497,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003974541171686239,
+      "loss": 1.8625,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039742772345951116,
+      "loss": 1.6466,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003974011945258778,
+      "loss": 1.5883,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003973745303858942,
+      "loss": 1.6464,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039734773105782353,
+      "loss": 1.7586,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039732079656002145,
+      "loss": 1.6374,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039729372691093634,
+      "loss": 1.7246,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000397266522129109,
+      "loss": 1.5848,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003972391822331729,
+      "loss": 1.8049,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039721170724185396,
+      "loss": 1.7783,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003971840971739707,
+      "loss": 1.5078,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003971563520484341,
+      "loss": 1.7298,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003971284718842477,
+      "loss": 1.8349,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003971004567005075,
+      "loss": 1.8207,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000397072306516402,
+      "loss": 1.713,
+      "step": 307
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039704402135121213,
+      "loss": 1.6281,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039701560122431135,
+      "loss": 1.6325,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039698704615516553,
+      "loss": 1.7483,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039695835616333276,
+      "loss": 1.489,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000396929531268464,
+      "loss": 1.7287,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039690057149030213,
+      "loss": 1.7216,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039687147684868264,
+      "loss": 1.7696,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003968422473635335,
+      "loss": 1.7033,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039681288305487485,
+      "loss": 1.677,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039678338394281916,
+      "loss": 1.5235,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039675375004757143,
+      "loss": 1.6583,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039672398138942874,
+      "loss": 1.5286,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039669407798878064,
+      "loss": 1.5743,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000396664039866109,
+      "loss": 1.6344,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003966338670419878,
+      "loss": 1.5803,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039660355953708337,
+      "loss": 1.6756,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039657311737215426,
+      "loss": 1.5601,
+      "step": 324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003965425405680513,
+      "loss": 1.6068,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003965118291457176,
+      "loss": 1.6917,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039648098312618826,
+      "loss": 1.5022,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039645000253059077,
+      "loss": 1.7106,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0003964188873801447,
+      "loss": 1.6934,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00039638763769616177,
+      "loss": 1.6792,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000396356253500046,
+      "loss": 1.5621,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039632473481329326,
+      "loss": 1.6579,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039629308165749184,
+      "loss": 1.6869,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003962612940543219,
+      "loss": 1.5827,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003962293720255558,
+      "loss": 1.7115,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000396197315593058,
+      "loss": 1.7004,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000396165124778785,
+      "loss": 1.673,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039613279960478516,
+      "loss": 1.5426,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003961003400931992,
+      "loss": 1.5172,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003960677462662594,
+      "loss": 1.7333,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003960350181462906,
+      "loss": 1.4797,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039600215575570923,
+      "loss": 1.7813,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003959691591170238,
+      "loss": 1.6425,
+      "step": 343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003959360282528347,
+      "loss": 1.5172,
+      "step": 344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003959027631858343,
+      "loss": 1.5473,
+      "step": 345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000395869363938807,
+      "loss": 1.5006,
+      "step": 346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003958358305346289,
+      "loss": 1.725,
+      "step": 347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003958021629962681,
+      "loss": 1.618,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003957683613467847,
+      "loss": 1.7454,
+      "step": 349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039573442560933035,
+      "loss": 1.4735,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003957003558071488,
+      "loss": 1.689,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039566615196357565,
+      "loss": 1.6993,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000395631814102038,
+      "loss": 1.5591,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000395597342246055,
+      "loss": 1.7427,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039556273641923757,
+      "loss": 1.7445,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039552799664528833,
+      "loss": 1.6865,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039549312294800157,
+      "loss": 1.5481,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039545811535126354,
+      "loss": 1.6455,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039542297387905193,
+      "loss": 1.6712,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003953876985554364,
+      "loss": 1.7702,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039535228940457796,
+      "loss": 1.8498,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003953167464507295,
+      "loss": 1.769,
+      "step": 362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003952810697182356,
+      "loss": 1.5027,
+      "step": 363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003952452592315324,
+      "loss": 1.5972,
+      "step": 364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039520931501514753,
+      "loss": 1.6795,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039517323709370043,
+      "loss": 1.7897,
+      "step": 366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039513702549190193,
+      "loss": 1.5261,
+      "step": 367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00039510068023455457,
+      "loss": 1.5994,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003950642013465523,
+      "loss": 1.6287,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003950275888528808,
+      "loss": 1.7588,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000394990842778617,
+      "loss": 1.6664,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039495396314892956,
+      "loss": 1.8034,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003949169499890784,
+      "loss": 1.6904,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039487980332441496,
+      "loss": 1.604,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039484252318038244,
+      "loss": 1.6644,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003948051095825149,
+      "loss": 1.7549,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039476756255643836,
+      "loss": 1.6544,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039472988212786974,
+      "loss": 1.6161,
+      "step": 378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039469206832261767,
+      "loss": 1.8024,
+      "step": 379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003946541211665821,
+      "loss": 1.655,
+      "step": 380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039461604068575406,
+      "loss": 1.5149,
+      "step": 381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039457782690621626,
+      "loss": 1.5712,
+      "step": 382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003945394798541424,
+      "loss": 1.7293,
+      "step": 383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039450099955579766,
+      "loss": 1.7491,
+      "step": 384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039446238603753846,
+      "loss": 1.7214,
+      "step": 385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039442363932581226,
+      "loss": 1.556,
+      "step": 386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003943847594471581,
+      "loss": 1.7184,
+      "step": 387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039434574642820597,
+      "loss": 1.7057,
+      "step": 388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003943066002956772,
+      "loss": 1.5008,
+      "step": 389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000394267321076384,
+      "loss": 1.6692,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003942279087972302,
+      "loss": 1.5715,
+      "step": 391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003941883634852104,
+      "loss": 1.6528,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039414868516741053,
+      "loss": 1.6049,
+      "step": 393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003941088738710074,
+      "loss": 1.5922,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039406892962326907,
+      "loss": 1.5338,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039402885245155475,
+      "loss": 1.5255,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003939886423833144,
+      "loss": 1.7363,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039394829944608927,
+      "loss": 1.6415,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039390782366751146,
+      "loss": 1.591,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003938672150753042,
+      "loss": 1.6248,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003938264736972815,
+      "loss": 1.684,
+      "step": 401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003937855995613485,
+      "loss": 1.6973,
+      "step": 402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039374459269550126,
+      "loss": 1.632,
+      "step": 403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039370345312782646,
+      "loss": 1.4909,
+      "step": 404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039366218088650215,
+      "loss": 1.6167,
+      "step": 405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003936207759997968,
+      "loss": 1.6814,
+      "step": 406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003935792384960701,
+      "loss": 1.7036,
+      "step": 407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0003935375684037723,
+      "loss": 1.5609,
+      "step": 408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00039349576575144457,
+      "loss": 1.5405,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003934538305677189,
+      "loss": 1.6685,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003934117628813181,
+      "loss": 1.7846,
+      "step": 411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003933695627210555,
+      "loss": 1.599,
+      "step": 412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003933272301158355,
+      "loss": 1.637,
+      "step": 413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003932847650946529,
+      "loss": 1.7403,
+      "step": 414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003932421676865935,
+      "loss": 1.6465,
+      "step": 415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003931994379208335,
+      "loss": 1.7376,
+      "step": 416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039315657582663987,
+      "loss": 1.5708,
+      "step": 417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003931135814333703,
+      "loss": 1.6891,
+      "step": 418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039307045477047294,
+      "loss": 1.5743,
+      "step": 419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003930271958674867,
+      "loss": 1.5618,
+      "step": 420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003929838047540408,
+      "loss": 1.5229,
+      "step": 421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039294028145985547,
+      "loss": 1.6628,
+      "step": 422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039289662601474097,
+      "loss": 1.6669,
+      "step": 423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003928528384485984,
+      "loss": 1.6269,
+      "step": 424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039280891879141916,
+      "loss": 1.7944,
+      "step": 425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003927648670732853,
+      "loss": 1.5369,
+      "step": 426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003927206833243693,
+      "loss": 1.6071,
+      "step": 427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003926763675749339,
+      "loss": 1.4575,
+      "step": 428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039263191985533237,
+      "loss": 1.7105,
+      "step": 429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003925873401960084,
+      "loss": 1.5246,
+      "step": 430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003925426286274961,
+      "loss": 1.5319,
+      "step": 431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003924977851804197,
+      "loss": 1.6269,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000392452809885494,
+      "loss": 1.6812,
+      "step": 433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039240770277352394,
+      "loss": 1.6768,
+      "step": 434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003923624638754048,
+      "loss": 1.734,
+      "step": 435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003923170932221223,
+      "loss": 1.6455,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000392271590844752,
+      "loss": 1.8286,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003922259567744602,
+      "loss": 1.6616,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039218019104250284,
+      "loss": 1.6312,
+      "step": 439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039213429368022655,
+      "loss": 1.7211,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039208826471906776,
+      "loss": 1.7451,
+      "step": 441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003920421041905532,
+      "loss": 1.558,
+      "step": 442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039199581212629966,
+      "loss": 1.8128,
+      "step": 443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003919493885580141,
+      "loss": 1.6611,
+      "step": 444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003919028335174935,
+      "loss": 1.5493,
+      "step": 445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003918561470366249,
+      "loss": 1.7107,
+      "step": 446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003918093291473852,
+      "loss": 1.6758,
+      "step": 447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0003917623798818417,
+      "loss": 1.7284,
+      "step": 448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003917152992721512,
+      "loss": 1.5784,
+      "step": 449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003916680873505609,
+      "loss": 1.6436,
+      "step": 450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039162074414940764,
+      "loss": 1.5409,
+      "step": 451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039157326970111836,
+      "loss": 1.549,
+      "step": 452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003915256640382097,
+      "loss": 1.7267,
+      "step": 453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003914779271932884,
+      "loss": 1.5716,
+      "step": 454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003914300591990509,
+      "loss": 1.6648,
+      "step": 455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003913820600882835,
+      "loss": 1.7487,
+      "step": 456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003913339298938623,
+      "loss": 1.4822,
+      "step": 457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039128566864875323,
+      "loss": 1.6741,
+      "step": 458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000391237276386012,
+      "loss": 1.6325,
+      "step": 459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003911887531387839,
+      "loss": 1.7415,
+      "step": 460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003911400989403042,
+      "loss": 1.6301,
+      "step": 461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003910913138238975,
+      "loss": 1.5065,
+      "step": 462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003910423978229784,
+      "loss": 1.6894,
+      "step": 463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003909933509710511,
+      "loss": 1.8452,
+      "step": 464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039094417330170914,
+      "loss": 1.551,
+      "step": 465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039089486484863604,
+      "loss": 1.5928,
+      "step": 466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039084542564560466,
+      "loss": 1.6541,
+      "step": 467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039079585572647746,
+      "loss": 1.5192,
+      "step": 468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003907461551252065,
+      "loss": 1.8088,
+      "step": 469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039069632387583327,
+      "loss": 1.6331,
+      "step": 470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003906463620124888,
+      "loss": 1.6655,
+      "step": 471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039059626956939355,
+      "loss": 1.5915,
+      "step": 472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003905460465808574,
+      "loss": 1.6448,
+      "step": 473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039049569308127974,
+      "loss": 1.5309,
+      "step": 474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039044520910514916,
+      "loss": 1.739,
+      "step": 475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039039459468704393,
+      "loss": 1.4387,
+      "step": 476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003903438498616313,
+      "loss": 1.6392,
+      "step": 477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000390292974663668,
+      "loss": 1.566,
+      "step": 478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039024196912800023,
+      "loss": 1.8407,
+      "step": 479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039019083328956317,
+      "loss": 1.803,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00039013956718338144,
+      "loss": 1.6142,
+      "step": 481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003900881708445688,
+      "loss": 1.5239,
+      "step": 482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003900366443083282,
+      "loss": 1.675,
+      "step": 483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003899849876099519,
+      "loss": 1.7271,
+      "step": 484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038993320078482114,
+      "loss": 1.6,
+      "step": 485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0003898812838684064,
+      "loss": 1.4937,
+      "step": 486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00038982923689626715,
+      "loss": 1.7205,
+      "step": 487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038977705990405216,
+      "loss": 1.5731,
+      "step": 488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003897247529274989,
+      "loss": 1.5392,
+      "step": 489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003896723160024343,
+      "loss": 1.5091,
+      "step": 490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038961974916477396,
+      "loss": 1.6609,
+      "step": 491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003895670524505226,
+      "loss": 1.5858,
+      "step": 492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038951422589577386,
+      "loss": 1.604,
+      "step": 493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003894612695367103,
+      "loss": 1.6486,
+      "step": 494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038940818340960347,
+      "loss": 1.6754,
+      "step": 495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038935496755081373,
+      "loss": 1.6549,
+      "step": 496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003893016219967903,
+      "loss": 1.6725,
+      "step": 497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003892481467840712,
+      "loss": 1.593,
+      "step": 498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003891945419492834,
+      "loss": 1.613,
+      "step": 499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038914080752914257,
+      "loss": 1.5978,
+      "step": 500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000389086943560453,
+      "loss": 1.758,
+      "step": 501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038903295008010783,
+      "loss": 1.7143,
+      "step": 502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038897882712508907,
+      "loss": 1.7088,
+      "step": 503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038892457473246704,
+      "loss": 1.7524,
+      "step": 504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003888701929394012,
+      "loss": 1.6681,
+      "step": 505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003888156817831391,
+      "loss": 1.7366,
+      "step": 506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038876104130101724,
+      "loss": 1.6069,
+      "step": 507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003887062715304607,
+      "loss": 1.7214,
+      "step": 508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000388651372508983,
+      "loss": 1.7612,
+      "step": 509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003885963442741862,
+      "loss": 1.6513,
+      "step": 510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038854118686376096,
+      "loss": 1.6339,
+      "step": 511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003884859003154862,
+      "loss": 1.5985,
+      "step": 512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003884304846672296,
+      "loss": 1.6362,
+      "step": 513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003883749399569469,
+      "loss": 1.8127,
+      "step": 514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038831926622268266,
+      "loss": 1.7264,
+      "step": 515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038826346350256946,
+      "loss": 1.7378,
+      "step": 516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038820753183482835,
+      "loss": 1.5021,
+      "step": 517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003881514712577687,
+      "loss": 1.653,
+      "step": 518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038809528180978826,
+      "loss": 1.5079,
+      "step": 519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003880389635293729,
+      "loss": 1.706,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003879825164550969,
+      "loss": 1.554,
+      "step": 521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003879259406256225,
+      "loss": 1.6374,
+      "step": 522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0003878692360797004,
+      "loss": 1.6435,
+      "step": 523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038781240285616923,
+      "loss": 1.7432,
+      "step": 524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000387755440993956,
+      "loss": 1.6771,
+      "step": 525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00038769835053207566,
+      "loss": 1.5618,
+      "step": 526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003876411315096313,
+      "loss": 1.654,
+      "step": 527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003875837839658139,
+      "loss": 1.6184,
+      "step": 528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003875263079399028,
+      "loss": 1.5148,
+      "step": 529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003874687034712651,
+      "loss": 1.5511,
+      "step": 530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038741097059935586,
+      "loss": 1.6583,
+      "step": 531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003873531093637182,
+      "loss": 1.6299,
+      "step": 532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003872951198039832,
+      "loss": 1.5602,
+      "step": 533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038723700195986973,
+      "loss": 1.8269,
+      "step": 534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038717875587118445,
+      "loss": 1.691,
+      "step": 535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000387120381577822,
+      "loss": 1.7941,
+      "step": 536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038706187911976474,
+      "loss": 1.6985,
+      "step": 537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000387003248537083,
+      "loss": 1.5676,
+      "step": 538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003869444898699346,
+      "loss": 1.6688,
+      "step": 539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003868856031585653,
+      "loss": 1.7335,
+      "step": 540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038682658844330846,
+      "loss": 1.6613,
+      "step": 541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000386767445764585,
+      "loss": 1.6753,
+      "step": 542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003867081751629039,
+      "loss": 1.5821,
+      "step": 543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038664877667886127,
+      "loss": 1.6122,
+      "step": 544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000386589250353141,
+      "loss": 1.5739,
+      "step": 545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003865295962265147,
+      "loss": 1.5059,
+      "step": 546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003864698143398413,
+      "loss": 1.7605,
+      "step": 547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003864099047340673,
+      "loss": 1.6589,
+      "step": 548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003863498674502267,
+      "loss": 1.7321,
+      "step": 549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000386289702529441,
+      "loss": 1.7304,
+      "step": 550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038622941001291894,
+      "loss": 1.6749,
+      "step": 551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038616898994195696,
+      "loss": 1.6343,
+      "step": 552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003861084423579385,
+      "loss": 1.6321,
+      "step": 553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038604776730233467,
+      "loss": 1.7723,
+      "step": 554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038598696481670364,
+      "loss": 1.5751,
+      "step": 555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003859260349426911,
+      "loss": 1.6355,
+      "step": 556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003858649777220296,
+      "loss": 1.4967,
+      "step": 557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003858037931965395,
+      "loss": 1.4851,
+      "step": 558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003857424814081278,
+      "loss": 1.5698,
+      "step": 559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038568104239878893,
+      "loss": 1.5247,
+      "step": 560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038561947621060445,
+      "loss": 1.6362,
+      "step": 561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000385557782885743,
+      "loss": 1.6348,
+      "step": 562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003854959624664603,
+      "loss": 1.6125,
+      "step": 563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0003854340149950991,
+      "loss": 1.6978,
+      "step": 564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00038537194051408926,
+      "loss": 1.6874,
+      "step": 565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003853097390659474,
+      "loss": 1.6011,
+      "step": 566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003852474106932775,
+      "loss": 1.6312,
+      "step": 567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038518495543877,
+      "loss": 1.7431,
+      "step": 568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038512237334520263,
+      "loss": 1.436,
+      "step": 569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003850596644554399,
+      "loss": 1.7691,
+      "step": 570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038499682881243294,
+      "loss": 1.5807,
+      "step": 571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038493386645922003,
+      "loss": 1.6818,
+      "step": 572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038487077743892605,
+      "loss": 1.5128,
+      "step": 573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003848075617947627,
+      "loss": 1.6996,
+      "step": 574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038474421957002835,
+      "loss": 1.644,
+      "step": 575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038468075080810805,
+      "loss": 1.6346,
+      "step": 576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003846171555524737,
+      "loss": 1.5746,
+      "step": 577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003845534338466836,
+      "loss": 1.6314,
+      "step": 578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038448958573438283,
+      "loss": 1.8287,
+      "step": 579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003844256112593029,
+      "loss": 1.5512,
+      "step": 580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003843615104652621,
+      "loss": 1.7027,
+      "step": 581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000384297283396165,
+      "loss": 1.5607,
+      "step": 582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003842329300960028,
+      "loss": 1.6542,
+      "step": 583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003841684506088531,
+      "loss": 1.5433,
+      "step": 584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038410384497888,
+      "loss": 1.5978,
+      "step": 585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038403911325033383,
+      "loss": 1.6236,
+      "step": 586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038397425546755155,
+      "loss": 1.6,
+      "step": 587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003839092716749563,
+      "loss": 1.574,
+      "step": 588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003838441619170574,
+      "loss": 1.4845,
+      "step": 589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003837789262384507,
+      "loss": 1.6633,
+      "step": 590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038371356468381825,
+      "loss": 1.653,
+      "step": 591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038364807729792815,
+      "loss": 1.5522,
+      "step": 592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003835824641256348,
+      "loss": 1.6637,
+      "step": 593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038351672521187875,
+      "loss": 1.6192,
+      "step": 594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003834508606016868,
+      "loss": 1.8062,
+      "step": 595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003833848703401715,
+      "loss": 1.5102,
+      "step": 596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003833187544725318,
+      "loss": 1.5951,
+      "step": 597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038325251304405255,
+      "loss": 1.5905,
+      "step": 598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003831861461001046,
+      "loss": 1.7055,
+      "step": 599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003831196536861448,
+      "loss": 1.6382,
+      "step": 600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00038305303584771586,
+      "loss": 1.6842,
+      "step": 601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003829862926304465,
+      "loss": 1.6128,
+      "step": 602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0003829194240800513,
+      "loss": 1.5902,
+      "step": 603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038285243024233057,
+      "loss": 1.603,
+      "step": 604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003827853111631706,
+      "loss": 1.6249,
+      "step": 605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003827180668885433,
+      "loss": 1.3987,
+      "step": 606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038265069746450646,
+      "loss": 1.5676,
+      "step": 607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003825832029372036,
+      "loss": 1.8549,
+      "step": 608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038251558335286373,
+      "loss": 1.5502,
+      "step": 609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038244783875780177,
+      "loss": 1.6703,
+      "step": 610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038237996919841805,
+      "loss": 1.7345,
+      "step": 611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038231197472119864,
+      "loss": 1.7552,
+      "step": 612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038224385537271514,
+      "loss": 1.6106,
+      "step": 613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003821756111996246,
+      "loss": 1.7377,
+      "step": 614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038210724224866966,
+      "loss": 1.5222,
+      "step": 615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003820387485666784,
+      "loss": 1.6635,
+      "step": 616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038197013020056423,
+      "loss": 1.4375,
+      "step": 617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038190138719732615,
+      "loss": 1.6819,
+      "step": 618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003818325196040484,
+      "loss": 1.6796,
+      "step": 619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003817635274679006,
+      "loss": 1.6895,
+      "step": 620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038169441083613765,
+      "loss": 1.7996,
+      "step": 621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003816251697560996,
+      "loss": 1.4623,
+      "step": 622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000381555804275212,
+      "loss": 1.6559,
+      "step": 623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038148631444098556,
+      "loss": 1.5815,
+      "step": 624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003814167003010159,
+      "loss": 1.5663,
+      "step": 625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000381346961902984,
+      "loss": 1.6197,
+      "step": 626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038127709929465595,
+      "loss": 1.5806,
+      "step": 627
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003812071125238828,
+      "loss": 1.648,
+      "step": 628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003811370016386008,
+      "loss": 1.5886,
+      "step": 629
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038106676668683105,
+      "loss": 1.4344,
+      "step": 630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038099640771667974,
+      "loss": 1.6881,
+      "step": 631
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000380925924776338,
+      "loss": 1.733,
+      "step": 632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003808553179140817,
+      "loss": 1.4853,
+      "step": 633
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038078458717827183,
+      "loss": 1.6185,
+      "step": 634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003807137326173541,
+      "loss": 1.6862,
+      "step": 635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003806427542798591,
+      "loss": 1.6541,
+      "step": 636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003805716522144021,
+      "loss": 1.5254,
+      "step": 637
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003805004264696831,
+      "loss": 1.5116,
+      "step": 638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000380429077094487,
+      "loss": 1.6079,
+      "step": 639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00038035760413768314,
+      "loss": 1.5912,
+      "step": 640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003802860076482257,
+      "loss": 1.6562,
+      "step": 641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0003802142876751533,
+      "loss": 1.7789,
+      "step": 642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003801424442675894,
+      "loss": 1.6394,
+      "step": 643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00038007047747474164,
+      "loss": 1.7994,
+      "step": 644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037999838734590247,
+      "loss": 1.7582,
+      "step": 645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037992617393044865,
+      "loss": 1.6606,
+      "step": 646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037985383727784155,
+      "loss": 1.8093,
+      "step": 647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003797813774376267,
+      "loss": 1.7377,
+      "step": 648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003797087944594343,
+      "loss": 1.6768,
+      "step": 649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003796360883929786,
+      "loss": 1.5083,
+      "step": 650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037956325928805833,
+      "loss": 1.7675,
+      "step": 651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037949030719455654,
+      "loss": 1.678,
+      "step": 652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003794172321624403,
+      "loss": 1.689,
+      "step": 653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003793440342417611,
+      "loss": 1.5533,
+      "step": 654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003792707134826546,
+      "loss": 1.6934,
+      "step": 655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003791972699353404,
+      "loss": 1.6117,
+      "step": 656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003791237036501223,
+      "loss": 1.6952,
+      "step": 657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003790500146773883,
+      "loss": 1.6012,
+      "step": 658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037897620306761025,
+      "loss": 1.6057,
+      "step": 659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037890226887134405,
+      "loss": 1.5526,
+      "step": 660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037882821213922963,
+      "loss": 1.5555,
+      "step": 661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037875403292199074,
+      "loss": 1.4534,
+      "step": 662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037867973127043514,
+      "loss": 1.6554,
+      "step": 663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003786053072354544,
+      "loss": 1.7754,
+      "step": 664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037853076086802383,
+      "loss": 1.8038,
+      "step": 665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003784560922192027,
+      "loss": 1.6227,
+      "step": 666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037838130134013384,
+      "loss": 1.5376,
+      "step": 667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037830638828204396,
+      "loss": 1.7412,
+      "step": 668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037823135309624336,
+      "loss": 1.5911,
+      "step": 669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003781561958341261,
+      "loss": 1.9199,
+      "step": 670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037808091654716963,
+      "loss": 1.5819,
+      "step": 671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037800551528693535,
+      "loss": 1.5266,
+      "step": 672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003779299921050678,
+      "loss": 1.4461,
+      "step": 673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037785434705329523,
+      "loss": 1.5974,
+      "step": 674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003777785801834294,
+      "loss": 1.5486,
+      "step": 675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003777026915473654,
+      "loss": 1.6846,
+      "step": 676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037762668119708175,
+      "loss": 1.4414,
+      "step": 677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003775505491846404,
+      "loss": 1.455,
+      "step": 678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003774742955621866,
+      "loss": 1.6806,
+      "step": 679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0003773979203819488,
+      "loss": 1.6142,
+      "step": 680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00037732142369623886,
+      "loss": 1.6237,
+      "step": 681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037724480555745174,
+      "loss": 1.6837,
+      "step": 682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003771680660180657,
+      "loss": 1.5569,
+      "step": 683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000377091205130642,
+      "loss": 1.7432,
+      "step": 684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037701422294782514,
+      "loss": 1.6136,
+      "step": 685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037693711952234275,
+      "loss": 1.6606,
+      "step": 686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037685989490700533,
+      "loss": 1.8267,
+      "step": 687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003767825491547065,
+      "loss": 1.6405,
+      "step": 688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037670508231842277,
+      "loss": 1.7604,
+      "step": 689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003766274944512137,
+      "loss": 1.7139,
+      "step": 690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003765497856062216,
+      "loss": 1.6443,
+      "step": 691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003764719558366719,
+      "loss": 1.6801,
+      "step": 692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037639400519587243,
+      "loss": 1.5054,
+      "step": 693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037631593373721434,
+      "loss": 1.6751,
+      "step": 694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037623774151417105,
+      "loss": 1.7075,
+      "step": 695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000376159428580299,
+      "loss": 1.6365,
+      "step": 696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003760809949892371,
+      "loss": 1.6795,
+      "step": 697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037600244079470717,
+      "loss": 1.754,
+      "step": 698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003759237660505133,
+      "loss": 1.8013,
+      "step": 699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037584497081054243,
+      "loss": 1.7363,
+      "step": 700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037576605512876395,
+      "loss": 1.5492,
+      "step": 701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003756870190592296,
+      "loss": 1.6954,
+      "step": 702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037560786265607377,
+      "loss": 1.5803,
+      "step": 703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003755285859735132,
+      "loss": 1.5493,
+      "step": 704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037544918906584694,
+      "loss": 1.6476,
+      "step": 705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003753696719874566,
+      "loss": 1.6574,
+      "step": 706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037529003479280586,
+      "loss": 1.7977,
+      "step": 707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003752102775364407,
+      "loss": 1.5634,
+      "step": 708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003751304002729895,
+      "loss": 1.7496,
+      "step": 709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003750504030571627,
+      "loss": 1.7538,
+      "step": 710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037497028594375297,
+      "loss": 1.5127,
+      "step": 711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000374890048987635,
+      "loss": 1.7468,
+      "step": 712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037480969224376566,
+      "loss": 1.7455,
+      "step": 713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003747292157671839,
+      "loss": 1.7747,
+      "step": 714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037464861961301047,
+      "loss": 1.5947,
+      "step": 715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037456790383644836,
+      "loss": 1.6351,
+      "step": 716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00037448706849278236,
+      "loss": 1.66,
+      "step": 717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003744061136373791,
+      "loss": 1.5671,
+      "step": 718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0003743250393256872,
+      "loss": 1.5424,
+      "step": 719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000374243845613237,
+      "loss": 1.6223,
+      "step": 720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037416253255564063,
+      "loss": 1.6559,
+      "step": 721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003740811002085921,
+      "loss": 1.6577,
+      "step": 722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003739995486278668,
+      "loss": 1.6025,
+      "step": 723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003739178778693222,
+      "loss": 1.5758,
+      "step": 724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037383608798889715,
+      "loss": 1.7676,
+      "step": 725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003737541790426121,
+      "loss": 1.7539,
+      "step": 726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037367215108656906,
+      "loss": 1.5668,
+      "step": 727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003735900041769516,
+      "loss": 1.668,
+      "step": 728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003735077383700248,
+      "loss": 1.8307,
+      "step": 729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003734253537221351,
+      "loss": 1.7394,
+      "step": 730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003733428502897104,
+      "loss": 1.8586,
+      "step": 731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037326022812925983,
+      "loss": 1.6289,
+      "step": 732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037317748729737404,
+      "loss": 1.7485,
+      "step": 733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037309462785072473,
+      "loss": 1.5752,
+      "step": 734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003730116498460651,
+      "loss": 1.5966,
+      "step": 735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037292855334022934,
+      "loss": 1.536,
+      "step": 736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003728453383901329,
+      "loss": 1.5516,
+      "step": 737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037276200505277233,
+      "loss": 1.7783,
+      "step": 738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003726785533852254,
+      "loss": 1.6644,
+      "step": 739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003725949834446506,
+      "loss": 1.5799,
+      "step": 740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003725112952882878,
+      "loss": 1.726,
+      "step": 741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003724274889734575,
+      "loss": 1.6901,
+      "step": 742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003723435645575616,
+      "loss": 1.9424,
+      "step": 743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003722595220980824,
+      "loss": 1.6959,
+      "step": 744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003721753616525832,
+      "loss": 1.5724,
+      "step": 745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003720910832787083,
+      "loss": 1.6969,
+      "step": 746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037200668703418256,
+      "loss": 1.6828,
+      "step": 747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037192217297681176,
+      "loss": 1.6534,
+      "step": 748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037183754116448207,
+      "loss": 1.6635,
+      "step": 749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037175279165516063,
+      "loss": 1.6784,
+      "step": 750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003716679245068951,
+      "loss": 1.4693,
+      "step": 751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037158293977781353,
+      "loss": 1.9277,
+      "step": 752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003714978375261248,
+      "loss": 1.8126,
+      "step": 753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003714126178101182,
+      "loss": 1.649,
+      "step": 754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003713272806881633,
+      "loss": 1.6612,
+      "step": 755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003712418262187103,
+      "loss": 1.626,
+      "step": 756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003711562544602895,
+      "loss": 1.7009,
+      "step": 757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00037107056547151194,
+      "loss": 1.4813,
+      "step": 758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0003709847593110686,
+      "loss": 1.6061,
+      "step": 759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037089883603773084,
+      "loss": 1.7086,
+      "step": 760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037081279571035035,
+      "loss": 1.6072,
+      "step": 761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037072663838785873,
+      "loss": 1.5301,
+      "step": 762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000370640364129268,
+      "loss": 1.7812,
+      "step": 763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037055397299367015,
+      "loss": 1.5838,
+      "step": 764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037046746504023703,
+      "loss": 1.656,
+      "step": 765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003703808403282209,
+      "loss": 1.4833,
+      "step": 766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037029409891695364,
+      "loss": 1.6497,
+      "step": 767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003702072408658473,
+      "loss": 1.7795,
+      "step": 768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037012026623439357,
+      "loss": 1.776,
+      "step": 769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00037003317508216434,
+      "loss": 1.5213,
+      "step": 770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003699459674688109,
+      "loss": 1.6342,
+      "step": 771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003698586434540647,
+      "loss": 1.6278,
+      "step": 772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003697712030977366,
+      "loss": 1.5861,
+      "step": 773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036968364645971735,
+      "loss": 1.6302,
+      "step": 774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036959597359997724,
+      "loss": 1.6089,
+      "step": 775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036950818457856624,
+      "loss": 1.7833,
+      "step": 776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036942027945561384,
+      "loss": 1.7146,
+      "step": 777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003693322582913289,
+      "loss": 1.4829,
+      "step": 778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003692441211460001,
+      "loss": 1.7925,
+      "step": 779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036915586807999534,
+      "loss": 1.7343,
+      "step": 780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003690674991537618,
+      "loss": 1.5837,
+      "step": 781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036897901442782636,
+      "loss": 1.6729,
+      "step": 782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036889041396279495,
+      "loss": 1.6381,
+      "step": 783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003688016978193528,
+      "loss": 1.6378,
+      "step": 784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003687128660582644,
+      "loss": 1.6915,
+      "step": 785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003686239187403736,
+      "loss": 1.8582,
+      "step": 786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036853485592660304,
+      "loss": 1.4633,
+      "step": 787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036844567767795484,
+      "loss": 1.5704,
+      "step": 788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003683563840555099,
+      "loss": 1.6925,
+      "step": 789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036826697512042827,
+      "loss": 1.6668,
+      "step": 790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036817745093394906,
+      "loss": 1.8282,
+      "step": 791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036808781155739015,
+      "loss": 1.607,
+      "step": 792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036799805705214846,
+      "loss": 1.6484,
+      "step": 793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003679081874796996,
+      "loss": 1.5758,
+      "step": 794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003678182029015982,
+      "loss": 1.6793,
+      "step": 795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0003677281033794775,
+      "loss": 1.5645,
+      "step": 796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00036763788897504953,
+      "loss": 1.6874,
+      "step": 797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000367547559750105,
+      "loss": 1.5802,
+      "step": 798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003674571157665133,
+      "loss": 1.5528,
+      "step": 799
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003673665570862224,
+      "loss": 1.5414,
+      "step": 800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003672758837712587,
+      "loss": 1.6786,
+      "step": 801
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003671850958837274,
+      "loss": 1.5255,
+      "step": 802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003670941934858119,
+      "loss": 1.5688,
+      "step": 803
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003670031766397742,
+      "loss": 1.5886,
+      "step": 804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003669120454079545,
+      "loss": 1.89,
+      "step": 805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036682079985277165,
+      "loss": 1.5677,
+      "step": 806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036672944003672255,
+      "loss": 1.6519,
+      "step": 807
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036663796602238243,
+      "loss": 1.7055,
+      "step": 808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036654637787240474,
+      "loss": 1.5883,
+      "step": 809
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036645467564952114,
+      "loss": 1.61,
+      "step": 810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003663628594165414,
+      "loss": 1.6775,
+      "step": 811
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003662709292363534,
+      "loss": 1.4352,
+      "step": 812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000366178885171923,
+      "loss": 1.5655,
+      "step": 813
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003660867272862941,
+      "loss": 1.5255,
+      "step": 814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036599445564258857,
+      "loss": 1.6361,
+      "step": 815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003659020703040062,
+      "loss": 1.4347,
+      "step": 816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003658095713338247,
+      "loss": 1.5647,
+      "step": 817
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003657169587953994,
+      "loss": 1.7996,
+      "step": 818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036562423275216373,
+      "loss": 1.5662,
+      "step": 819
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036553139326762863,
+      "loss": 1.7122,
+      "step": 820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003654384404053828,
+      "loss": 1.6529,
+      "step": 821
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036534537422909264,
+      "loss": 1.5574,
+      "step": 822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036525219480250214,
+      "loss": 1.6851,
+      "step": 823
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003651589021894328,
+      "loss": 1.7735,
+      "step": 824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036506549645378377,
+      "loss": 1.711,
+      "step": 825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003649719776595315,
+      "loss": 1.7488,
+      "step": 826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036487834587073007,
+      "loss": 1.6816,
+      "step": 827
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036478460115151084,
+      "loss": 1.7147,
+      "step": 828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003646907435660826,
+      "loss": 1.6348,
+      "step": 829
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003645967731787313,
+      "loss": 1.5697,
+      "step": 830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003645026900538203,
+      "loss": 1.756,
+      "step": 831
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003644084942557901,
+      "loss": 1.5116,
+      "step": 832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003643141858491584,
+      "loss": 1.6375,
+      "step": 833
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00036421976489852006,
+      "loss": 1.6181,
+      "step": 834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003641252314685469,
+      "loss": 1.4172,
+      "step": 835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000364030585623988,
+      "loss": 1.544,
+      "step": 836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0003639358274296692,
+      "loss": 1.6789,
+      "step": 837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036384095695049333,
+      "loss": 1.6391,
+      "step": 838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036374597425144044,
+      "loss": 1.6257,
+      "step": 839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000363650879397567,
+      "loss": 1.5339,
+      "step": 840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036355567245400654,
+      "loss": 1.7024,
+      "step": 841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036346035348596936,
+      "loss": 1.5804,
+      "step": 842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036336492255874237,
+      "loss": 1.4643,
+      "step": 843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003632693797376893,
+      "loss": 1.4877,
+      "step": 844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036317372508825056,
+      "loss": 1.4627,
+      "step": 845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003630779586759429,
+      "loss": 1.7036,
+      "step": 846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036298208056635976,
+      "loss": 1.6852,
+      "step": 847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036288609082517126,
+      "loss": 1.6766,
+      "step": 848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036278998951812377,
+      "loss": 1.5677,
+      "step": 849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003626937767110401,
+      "loss": 1.4498,
+      "step": 850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036259745246981935,
+      "loss": 1.6757,
+      "step": 851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003625010168604373,
+      "loss": 1.5938,
+      "step": 852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036240446994894557,
+      "loss": 1.7743,
+      "step": 853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003623078118014722,
+      "loss": 1.7202,
+      "step": 854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003622110424842216,
+      "loss": 1.627,
+      "step": 855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000362114162063474,
+      "loss": 1.568,
+      "step": 856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003620171706055859,
+      "loss": 1.676,
+      "step": 857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003619200681769899,
+      "loss": 1.6625,
+      "step": 858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036182285484419443,
+      "loss": 1.6611,
+      "step": 859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003617255306737841,
+      "loss": 1.8309,
+      "step": 860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036162809573241924,
+      "loss": 1.5852,
+      "step": 861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003615305500868362,
+      "loss": 1.7406,
+      "step": 862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036143289380384703,
+      "loss": 1.4455,
+      "step": 863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003613351269503397,
+      "loss": 1.601,
+      "step": 864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003612372495932777,
+      "loss": 1.7066,
+      "step": 865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003611392617997006,
+      "loss": 1.6644,
+      "step": 866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036104116363672303,
+      "loss": 1.5774,
+      "step": 867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003609429551715358,
+      "loss": 1.6126,
+      "step": 868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036084463647140485,
+      "loss": 1.6447,
+      "step": 869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036074620760367183,
+      "loss": 1.5225,
+      "step": 870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003606476686357538,
+      "loss": 1.541,
+      "step": 871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003605490196351433,
+      "loss": 1.4021,
+      "step": 872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000360450260669408,
+      "loss": 1.6662,
+      "step": 873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00036035139180619124,
+      "loss": 1.6792,
+      "step": 874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003602524131132113,
+      "loss": 1.5839,
+      "step": 875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0003601533246582619,
+      "loss": 1.8402,
+      "step": 876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003600541265092119,
+      "loss": 1.556,
+      "step": 877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003599548187340052,
+      "loss": 1.6817,
+      "step": 878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035985540140066083,
+      "loss": 1.7084,
+      "step": 879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035975587457727303,
+      "loss": 1.7476,
+      "step": 880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003596562383320107,
+      "loss": 1.6483,
+      "step": 881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035955649273311795,
+      "loss": 1.6566,
+      "step": 882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035945663784891374,
+      "loss": 1.6778,
+      "step": 883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003593566737477918,
+      "loss": 1.7335,
+      "step": 884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003592566004982207,
+      "loss": 1.6062,
+      "step": 885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035915641816874395,
+      "loss": 1.8055,
+      "step": 886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035905612682797945,
+      "loss": 1.7148,
+      "step": 887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003589557265446199,
+      "loss": 1.5053,
+      "step": 888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035885521738743275,
+      "loss": 1.5616,
+      "step": 889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035875459942525984,
+      "loss": 1.6662,
+      "step": 890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003586538727270177,
+      "loss": 1.6985,
+      "step": 891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003585530373616972,
+      "loss": 1.8217,
+      "step": 892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003584520933983636,
+      "loss": 1.4381,
+      "step": 893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035835104090615674,
+      "loss": 1.6774,
+      "step": 894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003582498799542907,
+      "loss": 1.5872,
+      "step": 895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035814861061205373,
+      "loss": 1.5814,
+      "step": 896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003580472329488084,
+      "loss": 1.7271,
+      "step": 897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003579457470339917,
+      "loss": 1.7893,
+      "step": 898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003578441529371143,
+      "loss": 1.6335,
+      "step": 899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035774245072776146,
+      "loss": 1.4715,
+      "step": 900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000357640640475592,
+      "loss": 1.5212,
+      "step": 901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035753872225033925,
+      "loss": 1.4646,
+      "step": 902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035743669612181003,
+      "loss": 1.6283,
+      "step": 903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035733456215988543,
+      "loss": 1.5013,
+      "step": 904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003572323204345202,
+      "loss": 1.6665,
+      "step": 905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003571299710157429,
+      "loss": 1.6076,
+      "step": 906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003570275139736559,
+      "loss": 1.8144,
+      "step": 907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035692494937843535,
+      "loss": 1.6256,
+      "step": 908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000356822277300331,
+      "loss": 1.7083,
+      "step": 909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003567194978096661,
+      "loss": 1.6267,
+      "step": 910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003566166109768378,
+      "loss": 1.6438,
+      "step": 911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035651361687231634,
+      "loss": 1.5485,
+      "step": 912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003564105155666458,
+      "loss": 1.4371,
+      "step": 913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0003563073071304434,
+      "loss": 1.5437,
+      "step": 914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00035620399163440006,
+      "loss": 1.5017,
+      "step": 915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003561005691492797,
+      "loss": 1.6389,
+      "step": 916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003559970397459197,
+      "loss": 1.464,
+      "step": 917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035589340349523075,
+      "loss": 1.6925,
+      "step": 918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003557896604681965,
+      "loss": 1.7354,
+      "step": 919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035568581073587373,
+      "loss": 1.5807,
+      "step": 920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035558185436939267,
+      "loss": 1.638,
+      "step": 921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003554777914399562,
+      "loss": 1.4431,
+      "step": 922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003553736220188403,
+      "loss": 1.4802,
+      "step": 923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000355269346177394,
+      "loss": 1.6591,
+      "step": 924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003551649639870391,
+      "loss": 1.7172,
+      "step": 925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003550604755192703,
+      "loss": 1.456,
+      "step": 926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003549558808456551,
+      "loss": 1.7723,
+      "step": 927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003548511800378336,
+      "loss": 1.7289,
+      "step": 928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035474637316751877,
+      "loss": 1.7105,
+      "step": 929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003546414603064962,
+      "loss": 1.6791,
+      "step": 930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035453644152662393,
+      "loss": 1.6817,
+      "step": 931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035443131689983287,
+      "loss": 1.7588,
+      "step": 932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000354326086498126,
+      "loss": 1.6242,
+      "step": 933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003542207503935791,
+      "loss": 1.6501,
+      "step": 934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003541153086583402,
+      "loss": 1.6213,
+      "step": 935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003540097613646296,
+      "loss": 1.7623,
+      "step": 936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035390410858474015,
+      "loss": 1.6942,
+      "step": 937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035379835039103666,
+      "loss": 1.7283,
+      "step": 938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003536924868559563,
+      "loss": 1.7197,
+      "step": 939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003535865180520084,
+      "loss": 1.5329,
+      "step": 940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035348044405177434,
+      "loss": 1.6563,
+      "step": 941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003533742649279076,
+      "loss": 1.5709,
+      "step": 942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003532679807531336,
+      "loss": 1.7297,
+      "step": 943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035316159160024986,
+      "loss": 1.5328,
+      "step": 944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003530550975421255,
+      "loss": 1.6089,
+      "step": 945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035294849865170186,
+      "loss": 1.5144,
+      "step": 946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003528417950019918,
+      "loss": 1.6538,
+      "step": 947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035273498666608006,
+      "loss": 1.6211,
+      "step": 948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003526280737171231,
+      "loss": 1.5184,
+      "step": 949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035252105622834897,
+      "loss": 1.6793,
+      "step": 950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0003524139342730573,
+      "loss": 1.6752,
+      "step": 951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035230670792461946,
+      "loss": 1.5023,
+      "step": 952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035219937725647804,
+      "loss": 1.6061,
+      "step": 953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00035209194234214725,
+      "loss": 1.6007,
+      "step": 954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035198440325521276,
+      "loss": 1.6313,
+      "step": 955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003518767600693314,
+      "loss": 1.6372,
+      "step": 956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003517690128582315,
+      "loss": 1.6819,
+      "step": 957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003516611616957125,
+      "loss": 1.6247,
+      "step": 958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003515532066556451,
+      "loss": 1.4948,
+      "step": 959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035144514781197114,
+      "loss": 1.5885,
+      "step": 960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035133698523870355,
+      "loss": 1.531,
+      "step": 961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035122871900992634,
+      "loss": 1.6559,
+      "step": 962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003511203491997944,
+      "loss": 1.6803,
+      "step": 963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035101187588253374,
+      "loss": 1.6133,
+      "step": 964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035090329913244116,
+      "loss": 1.5852,
+      "step": 965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003507946190238843,
+      "loss": 1.6007,
+      "step": 966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035068583563130167,
+      "loss": 1.6832,
+      "step": 967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035057694902920246,
+      "loss": 1.6597,
+      "step": 968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003504679592921666,
+      "loss": 1.664,
+      "step": 969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003503588664948445,
+      "loss": 1.5242,
+      "step": 970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035024967071195734,
+      "loss": 1.7077,
+      "step": 971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003501403720182969,
+      "loss": 1.6798,
+      "step": 972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00035003097048872517,
+      "loss": 1.71,
+      "step": 973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034992146619817474,
+      "loss": 1.5513,
+      "step": 974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003498118592216487,
+      "loss": 1.7077,
+      "step": 975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034970214963422024,
+      "loss": 1.7366,
+      "step": 976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000349592337511033,
+      "loss": 1.7018,
+      "step": 977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034948242292730074,
+      "loss": 1.7332,
+      "step": 978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003493724059583075,
+      "loss": 1.8433,
+      "step": 979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034926228667940746,
+      "loss": 1.6216,
+      "step": 980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034915206516602463,
+      "loss": 1.6092,
+      "step": 981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003490417414936534,
+      "loss": 1.722,
+      "step": 982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034893131573785787,
+      "loss": 1.5727,
+      "step": 983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034882078797427216,
+      "loss": 1.5996,
+      "step": 984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034871015827860026,
+      "loss": 1.5437,
+      "step": 985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034859942672661596,
+      "loss": 1.6642,
+      "step": 986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003484885933941628,
+      "loss": 1.5446,
+      "step": 987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034837765835715414,
+      "loss": 1.4511,
+      "step": 988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034826662169157275,
+      "loss": 1.5277,
+      "step": 989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034815548347347133,
+      "loss": 1.7051,
+      "step": 990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003480442437789718,
+      "loss": 1.8108,
+      "step": 991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00034793290268426595,
+      "loss": 1.8663,
+      "step": 992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0003478214602656148,
+      "loss": 1.6647,
+      "step": 993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003477099165993487,
+      "loss": 1.7646,
+      "step": 994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003475982717618675,
+      "loss": 1.5646,
+      "step": 995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034748652582964034,
+      "loss": 1.5051,
+      "step": 996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003473746788792055,
+      "loss": 1.7521,
+      "step": 997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034726273098717057,
+      "loss": 1.7792,
+      "step": 998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003471506822302121,
+      "loss": 1.6175,
+      "step": 999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000347038532685076,
+      "loss": 1.6643,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003469262824285769,
+      "loss": 1.613,
+      "step": 1001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003468139315375987,
+      "loss": 1.6862,
+      "step": 1002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000346701480089094,
+      "loss": 1.6329,
+      "step": 1003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003465889281600845,
+      "loss": 1.5296,
+      "step": 1004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003464762758276605,
+      "loss": 1.6299,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003463635231689811,
+      "loss": 1.5426,
+      "step": 1006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034625067026127426,
+      "loss": 1.5857,
+      "step": 1007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034613771718183654,
+      "loss": 1.7032,
+      "step": 1008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003460246640080331,
+      "loss": 1.7083,
+      "step": 1009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034591151081729755,
+      "loss": 1.6932,
+      "step": 1010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034579825768713213,
+      "loss": 1.6534,
+      "step": 1011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034568490469510764,
+      "loss": 1.5892,
+      "step": 1012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034557145191886293,
+      "loss": 1.6582,
+      "step": 1013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034545789943610564,
+      "loss": 1.5769,
+      "step": 1014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034534424732461134,
+      "loss": 1.5182,
+      "step": 1015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000345230495662224,
+      "loss": 1.6186,
+      "step": 1016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003451166445268558,
+      "loss": 1.745,
+      "step": 1017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003450026939964868,
+      "loss": 1.3753,
+      "step": 1018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034488864414916555,
+      "loss": 1.5549,
+      "step": 1019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034477449506300844,
+      "loss": 1.5743,
+      "step": 1020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034466024681619964,
+      "loss": 1.5706,
+      "step": 1021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003445458994869915,
+      "loss": 1.4817,
+      "step": 1022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003444314531537041,
+      "loss": 1.7491,
+      "step": 1023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034431690789472533,
+      "loss": 1.538,
+      "step": 1024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034420226378851105,
+      "loss": 1.497,
+      "step": 1025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003440875209135845,
+      "loss": 1.6419,
+      "step": 1026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034397267934853674,
+      "loss": 1.6232,
+      "step": 1027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003438577391720264,
+      "loss": 1.4944,
+      "step": 1028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00034374270046277965,
+      "loss": 1.4733,
+      "step": 1029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003436275632995903,
+      "loss": 1.703,
+      "step": 1030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003435123277613193,
+      "loss": 1.6112,
+      "step": 1031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0003433969939268951,
+      "loss": 1.6088,
+      "step": 1032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034328156187531366,
+      "loss": 1.6745,
+      "step": 1033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034316603168563795,
+      "loss": 1.6352,
+      "step": 1034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034305040343699834,
+      "loss": 1.7108,
+      "step": 1035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034293467720859224,
+      "loss": 1.5367,
+      "step": 1036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003428188530796842,
+      "loss": 1.7411,
+      "step": 1037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034270293112960583,
+      "loss": 1.6243,
+      "step": 1038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034258691143775586,
+      "loss": 1.8486,
+      "step": 1039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003424707940835998,
+      "loss": 1.4878,
+      "step": 1040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034235457914667014,
+      "loss": 1.8384,
+      "step": 1041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003422382667065662,
+      "loss": 1.7218,
+      "step": 1042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000342121856842954,
+      "loss": 1.5259,
+      "step": 1043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003420053496355663,
+      "loss": 1.7574,
+      "step": 1044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003418887451642028,
+      "loss": 1.6499,
+      "step": 1045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034177204350872934,
+      "loss": 1.6151,
+      "step": 1046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003416552447490788,
+      "loss": 1.5587,
+      "step": 1047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034153834896525037,
+      "loss": 1.564,
+      "step": 1048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003414213562373095,
+      "loss": 1.5364,
+      "step": 1049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003413042666453884,
+      "loss": 1.6243,
+      "step": 1050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034118708026968533,
+      "loss": 1.764,
+      "step": 1051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003410697971904651,
+      "loss": 1.6464,
+      "step": 1052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003409524174880585,
+      "loss": 1.5382,
+      "step": 1053
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034083494124286266,
+      "loss": 1.5711,
+      "step": 1054
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003407173685353408,
+      "loss": 1.668,
+      "step": 1055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003405996994460221,
+      "loss": 1.4977,
+      "step": 1056
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00034048193405550206,
+      "loss": 1.5744,
+      "step": 1057
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003403640724444418,
+      "loss": 1.5391,
+      "step": 1058
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003402461146935685,
+      "loss": 1.6653,
+      "step": 1059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003401280608836752,
+      "loss": 1.6873,
+      "step": 1060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003400099110956206,
+      "loss": 1.8143,
+      "step": 1061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003398916654103294,
+      "loss": 1.6314,
+      "step": 1062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003397733239087918,
+      "loss": 1.5244,
+      "step": 1063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033965488667206357,
+      "loss": 1.5605,
+      "step": 1064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003395363537812662,
+      "loss": 1.6605,
+      "step": 1065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033941772531758654,
+      "loss": 1.5791,
+      "step": 1066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003392990013622771,
+      "loss": 1.6512,
+      "step": 1067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003391801819966557,
+      "loss": 1.4507,
+      "step": 1068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003390612673021054,
+      "loss": 1.3856,
+      "step": 1069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0003389422573600747,
+      "loss": 1.476,
+      "step": 1070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00033882315225207735,
+      "loss": 1.4502,
+      "step": 1071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003387039520596921,
+      "loss": 1.4469,
+      "step": 1072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033858465686456313,
+      "loss": 1.6909,
+      "step": 1073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003384652667483993,
+      "loss": 1.6034,
+      "step": 1074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003383457817929748,
+      "loss": 1.6907,
+      "step": 1075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033822620208012863,
+      "loss": 1.4227,
+      "step": 1076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033810652769176487,
+      "loss": 1.703,
+      "step": 1077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033798675870985213,
+      "loss": 1.4151,
+      "step": 1078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033786689521642414,
+      "loss": 1.8087,
+      "step": 1079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003377469372935791,
+      "loss": 1.6159,
+      "step": 1080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033762688502348006,
+      "loss": 1.5592,
+      "step": 1081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003375067384883546,
+      "loss": 1.5487,
+      "step": 1082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003373864977704949,
+      "loss": 1.6412,
+      "step": 1083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033726616295225777,
+      "loss": 1.5919,
+      "step": 1084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033714573411606423,
+      "loss": 1.5356,
+      "step": 1085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003370252113443998,
+      "loss": 1.5686,
+      "step": 1086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033690459471981437,
+      "loss": 1.5398,
+      "step": 1087
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033678388432492216,
+      "loss": 1.5646,
+      "step": 1088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003366630802424015,
+      "loss": 1.7746,
+      "step": 1089
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000336542182554995,
+      "loss": 1.5381,
+      "step": 1090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003364211913455092,
+      "loss": 1.6791,
+      "step": 1091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000336300106696815,
+      "loss": 1.6803,
+      "step": 1092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033617892869184697,
+      "loss": 1.5346,
+      "step": 1093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003360576574136038,
+      "loss": 1.6946,
+      "step": 1094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033593629294514823,
+      "loss": 1.5399,
+      "step": 1095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003358148353696064,
+      "loss": 1.5099,
+      "step": 1096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033569328477016857,
+      "loss": 1.4843,
+      "step": 1097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003355716412300886,
+      "loss": 1.6703,
+      "step": 1098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000335449904832684,
+      "loss": 1.6933,
+      "step": 1099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033532807566133595,
+      "loss": 1.6817,
+      "step": 1100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000335206153799489,
+      "loss": 1.631,
+      "step": 1101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033508413933065143,
+      "loss": 1.7545,
+      "step": 1102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003349620323383948,
+      "loss": 1.5482,
+      "step": 1103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000334839832906354,
+      "loss": 1.5616,
+      "step": 1104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003347175411182274,
+      "loss": 1.6343,
+      "step": 1105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033459515705777643,
+      "loss": 1.5963,
+      "step": 1106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033447268080882586,
+      "loss": 1.5286,
+      "step": 1107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033435011245526366,
+      "loss": 1.5815,
+      "step": 1108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0003342274520810406,
+      "loss": 1.5509,
+      "step": 1109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00033410469977017084,
+      "loss": 1.5919,
+      "step": 1110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003339818556067313,
+      "loss": 1.523,
+      "step": 1111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033385891967486185,
+      "loss": 1.5834,
+      "step": 1112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033373589205876506,
+      "loss": 1.3742,
+      "step": 1113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003336127728427067,
+      "loss": 1.5829,
+      "step": 1114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033348956211101487,
+      "loss": 1.7769,
+      "step": 1115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003333662599480806,
+      "loss": 1.5555,
+      "step": 1116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003332428664383574,
+      "loss": 1.5663,
+      "step": 1117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033311938166636143,
+      "loss": 1.682,
+      "step": 1118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033299580571667127,
+      "loss": 1.6078,
+      "step": 1119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003328721386739282,
+      "loss": 1.5645,
+      "step": 1120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003327483806228355,
+      "loss": 1.6554,
+      "step": 1121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033262453164815904,
+      "loss": 1.647,
+      "step": 1122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033250059183472706,
+      "loss": 1.6142,
+      "step": 1123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033237656126742967,
+      "loss": 1.3371,
+      "step": 1124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033225244003121947,
+      "loss": 1.7582,
+      "step": 1125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033212822821111104,
+      "loss": 1.5812,
+      "step": 1126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003320039258921809,
+      "loss": 1.6319,
+      "step": 1127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003318795331595678,
+      "loss": 1.6679,
+      "step": 1128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003317550500984722,
+      "loss": 1.5731,
+      "step": 1129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003316304767941564,
+      "loss": 1.7698,
+      "step": 1130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033150581333194493,
+      "loss": 1.6733,
+      "step": 1131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003313810597972234,
+      "loss": 1.4349,
+      "step": 1132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033125621627543966,
+      "loss": 1.5253,
+      "step": 1133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000331131282852103,
+      "loss": 1.7064,
+      "step": 1134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003310062596127842,
+      "loss": 1.6397,
+      "step": 1135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003308811466431157,
+      "loss": 1.8502,
+      "step": 1136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003307559440287914,
+      "loss": 1.5914,
+      "step": 1137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033063065185556654,
+      "loss": 1.6453,
+      "step": 1138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003305052702092576,
+      "loss": 1.5936,
+      "step": 1139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003303797991757425,
+      "loss": 1.5098,
+      "step": 1140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00033025423884096045,
+      "loss": 1.5962,
+      "step": 1141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003301285892909116,
+      "loss": 1.6867,
+      "step": 1142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003300028506116573,
+      "loss": 1.5216,
+      "step": 1143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032987702288932,
+      "loss": 1.6307,
+      "step": 1144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032975110621008324,
+      "loss": 1.7512,
+      "step": 1145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003296251006601911,
+      "loss": 1.607,
+      "step": 1146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032949900632594896,
+      "loss": 1.6309,
+      "step": 1147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00032937282329372283,
+      "loss": 1.6502,
+      "step": 1148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003292465516499394,
+      "loss": 1.5231,
+      "step": 1149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032912019148108616,
+      "loss": 1.5893,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032899374287371125,
+      "loss": 1.5379,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032886720591442326,
+      "loss": 1.5696,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032874058068989146,
+      "loss": 1.5037,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032861386728684543,
+      "loss": 1.608,
+      "step": 1154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032848706579207527,
+      "loss": 1.5589,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003283601762924312,
+      "loss": 1.5463,
+      "step": 1156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003282331988748241,
+      "loss": 1.5777,
+      "step": 1157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003281061336262246,
+      "loss": 1.6662,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003279789806336639,
+      "loss": 1.5816,
+      "step": 1159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003278517399842331,
+      "loss": 1.5934,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032772441176508323,
+      "loss": 1.641,
+      "step": 1161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003275969960634256,
+      "loss": 1.6322,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003274694929665312,
+      "loss": 1.5589,
+      "step": 1163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003273419025617309,
+      "loss": 1.6211,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003272142249364155,
+      "loss": 1.734,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003270864601780355,
+      "loss": 1.7946,
+      "step": 1166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003269586083741009,
+      "loss": 1.4866,
+      "step": 1167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032683066961218163,
+      "loss": 1.5883,
+      "step": 1168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003267026439799069,
+      "loss": 1.6302,
+      "step": 1169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003265745315649656,
+      "loss": 1.461,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032644633245510586,
+      "loss": 1.7097,
+      "step": 1171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032631804673813547,
+      "loss": 1.4899,
+      "step": 1172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003261896745019214,
+      "loss": 1.5518,
+      "step": 1173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032606121583438974,
+      "loss": 1.5662,
+      "step": 1174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000325932670823526,
+      "loss": 1.5803,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003258040395573747,
+      "loss": 1.6027,
+      "step": 1176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032567532212403956,
+      "loss": 1.636,
+      "step": 1177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003255465186116832,
+      "loss": 1.6611,
+      "step": 1178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003254176291085272,
+      "loss": 1.5739,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032528865370285207,
+      "loss": 1.6826,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003251595924829973,
+      "loss": 1.542,
+      "step": 1181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003250304455373608,
+      "loss": 1.4313,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032490121295439966,
+      "loss": 1.7157,
+      "step": 1183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003247718948226292,
+      "loss": 1.4181,
+      "step": 1184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00032464249123062355,
+      "loss": 1.6829,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003245130022670154,
+      "loss": 1.5138,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003243834280204958,
+      "loss": 1.4924,
+      "step": 1187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003242537685798143,
+      "loss": 1.6462,
+      "step": 1188
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032412402403377883,
+      "loss": 1.5998,
+      "step": 1189
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003239941944712554,
+      "loss": 1.7109,
+      "step": 1190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003238642799811685,
+      "loss": 1.5108,
+      "step": 1191
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003237342806525007,
+      "loss": 1.6894,
+      "step": 1192
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032360419657429266,
+      "loss": 1.7361,
+      "step": 1193
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003234740278356431,
+      "loss": 1.7535,
+      "step": 1194
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003233437745257087,
+      "loss": 1.5421,
+      "step": 1195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000323213436733704,
+      "loss": 1.7633,
+      "step": 1196
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003230830145489017,
+      "loss": 1.5741,
+      "step": 1197
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032295250806063197,
+      "loss": 1.6461,
+      "step": 1198
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032282191735828286,
+      "loss": 1.5867,
+      "step": 1199
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032269124253130017,
+      "loss": 1.6033,
+      "step": 1200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003225604836691871,
+      "loss": 1.5716,
+      "step": 1201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032242964086150465,
+      "loss": 1.6491,
+      "step": 1202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032229871419787125,
+      "loss": 1.6562,
+      "step": 1203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032216770376796263,
+      "loss": 1.5501,
+      "step": 1204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003220366096615121,
+      "loss": 1.7118,
+      "step": 1205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032190543196831007,
+      "loss": 1.6797,
+      "step": 1206
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032177417077820434,
+      "loss": 1.5592,
+      "step": 1207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032164282618109995,
+      "loss": 1.63,
+      "step": 1208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032151139826695893,
+      "loss": 1.5523,
+      "step": 1209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032137988712580034,
+      "loss": 1.554,
+      "step": 1210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003212482928477005,
+      "loss": 1.5144,
+      "step": 1211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032111661552279244,
+      "loss": 1.6007,
+      "step": 1212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032098485524126617,
+      "loss": 1.6295,
+      "step": 1213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003208530120933684,
+      "loss": 1.7251,
+      "step": 1214
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003207210861694028,
+      "loss": 1.4659,
+      "step": 1215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003205890775597296,
+      "loss": 1.7953,
+      "step": 1216
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003204569863547657,
+      "loss": 1.5445,
+      "step": 1217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003203248126449845,
+      "loss": 1.6421,
+      "step": 1218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032019255652091606,
+      "loss": 1.5815,
+      "step": 1219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003200602180731467,
+      "loss": 1.6216,
+      "step": 1220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003199277973923194,
+      "loss": 1.5336,
+      "step": 1221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003197952945691331,
+      "loss": 1.494,
+      "step": 1222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031966270969434324,
+      "loss": 1.5711,
+      "step": 1223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003195300428587615,
+      "loss": 1.589,
+      "step": 1224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003193972941532556,
+      "loss": 1.5672,
+      "step": 1225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031926446366874916,
+      "loss": 1.5589,
+      "step": 1226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031913155149622223,
+      "loss": 1.5723,
+      "step": 1227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031899855772671043,
+      "loss": 1.4558,
+      "step": 1228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031886548245130557,
+      "loss": 1.543,
+      "step": 1229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031873232576115504,
+      "loss": 1.7061,
+      "step": 1230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031859908774746205,
+      "loss": 1.6974,
+      "step": 1231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003184657685014856,
+      "loss": 1.6647,
+      "step": 1232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003183323681145402,
+      "loss": 1.5667,
+      "step": 1233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003181988866779962,
+      "loss": 1.4718,
+      "step": 1234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003180653242832791,
+      "loss": 1.6761,
+      "step": 1235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003179316810218701,
+      "loss": 1.5363,
+      "step": 1236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003177979569853057,
+      "loss": 1.6697,
+      "step": 1237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003176641522651778,
+      "loss": 1.7792,
+      "step": 1238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003175302669531333,
+      "loss": 1.6021,
+      "step": 1239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031739630114087483,
+      "loss": 1.6237,
+      "step": 1240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003172622549201596,
+      "loss": 1.3089,
+      "step": 1241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003171281283828001,
+      "loss": 1.6997,
+      "step": 1242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000316993921620664,
+      "loss": 1.6001,
+      "step": 1243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003168596347256737,
+      "loss": 1.7336,
+      "step": 1244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031672526778980657,
+      "loss": 1.6224,
+      "step": 1245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003165908209050948,
+      "loss": 1.5767,
+      "step": 1246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003164562941636254,
+      "loss": 1.6763,
+      "step": 1247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031632168765753985,
+      "loss": 1.5773,
+      "step": 1248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031618700147903456,
+      "loss": 1.5524,
+      "step": 1249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031605223572036027,
+      "loss": 1.7016,
+      "step": 1250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003159173904738224,
+      "loss": 1.642,
+      "step": 1251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003157824658317808,
+      "loss": 1.4636,
+      "step": 1252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003156474618866495,
+      "loss": 1.6121,
+      "step": 1253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031551237873089706,
+      "loss": 1.5131,
+      "step": 1254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031537721645704634,
+      "loss": 1.6786,
+      "step": 1255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003152419751576741,
+      "loss": 1.6942,
+      "step": 1256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031510665492541143,
+      "loss": 1.5515,
+      "step": 1257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031497125585294353,
+      "loss": 1.5231,
+      "step": 1258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031483577803300946,
+      "loss": 1.7488,
+      "step": 1259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003147002215584023,
+      "loss": 1.4791,
+      "step": 1260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031456458652196915,
+      "loss": 1.7005,
+      "step": 1261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003144288730166105,
+      "loss": 1.7452,
+      "step": 1262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000314293081135281,
+      "loss": 1.8606,
+      "step": 1263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003141572109709887,
+      "loss": 1.6484,
+      "step": 1264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003140212626167955,
+      "loss": 1.7995,
+      "step": 1265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003138852361658167,
+      "loss": 1.604,
+      "step": 1266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003137491317112212,
+      "loss": 1.7919,
+      "step": 1267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031361294934623114,
+      "loss": 1.6045,
+      "step": 1268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031347668916412234,
+      "loss": 1.7168,
+      "step": 1269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003133403512582235,
+      "loss": 1.5723,
+      "step": 1270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031320393572191696,
+      "loss": 1.589,
+      "step": 1271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031306744264863806,
+      "loss": 1.4084,
+      "step": 1272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003129308721318751,
+      "loss": 1.6013,
+      "step": 1273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003127942242651696,
+      "loss": 1.6407,
+      "step": 1274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003126574991421162,
+      "loss": 1.6853,
+      "step": 1275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000312520696856362,
+      "loss": 1.4918,
+      "step": 1276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031238381750160754,
+      "loss": 1.6117,
+      "step": 1277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031224686117160556,
+      "loss": 1.5732,
+      "step": 1278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000312109827960162,
+      "loss": 1.6469,
+      "step": 1279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003119727179611351,
+      "loss": 1.5207,
+      "step": 1280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031183553126843603,
+      "loss": 1.6142,
+      "step": 1281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031169826797602815,
+      "loss": 1.666,
+      "step": 1282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031156092817792753,
+      "loss": 1.6464,
+      "step": 1283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003114235119682026,
+      "loss": 1.7152,
+      "step": 1284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000311286019440974,
+      "loss": 1.6024,
+      "step": 1285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003111484506904148,
+      "loss": 1.5294,
+      "step": 1286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003110108058107502,
+      "loss": 1.839,
+      "step": 1287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003108730848962575,
+      "loss": 1.5395,
+      "step": 1288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031073528804126616,
+      "loss": 1.5586,
+      "step": 1289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003105974153401577,
+      "loss": 1.6043,
+      "step": 1290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003104594668873653,
+      "loss": 1.621,
+      "step": 1291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003103214427773745,
+      "loss": 1.643,
+      "step": 1292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00031018334310472227,
+      "loss": 1.5992,
+      "step": 1293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003100451679639974,
+      "loss": 1.6329,
+      "step": 1294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003099069174498406,
+      "loss": 1.4865,
+      "step": 1295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030976859165694385,
+      "loss": 1.6112,
+      "step": 1296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003096301906800511,
+      "loss": 1.4441,
+      "step": 1297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003094917146139573,
+      "loss": 1.6196,
+      "step": 1298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030935316355350934,
+      "loss": 1.6713,
+      "step": 1299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003092145375936052,
+      "loss": 1.6121,
+      "step": 1300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003090758368291941,
+      "loss": 1.6813,
+      "step": 1301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003089370613552767,
+      "loss": 1.6353,
+      "step": 1302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003087982112669047,
+      "loss": 1.4749,
+      "step": 1303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00030865928665918096,
+      "loss": 1.5434,
+      "step": 1304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0003085202876272593,
+      "loss": 1.7776,
+      "step": 1305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030838121426634467,
+      "loss": 1.6476,
+      "step": 1306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003082420666716927,
+      "loss": 1.5497,
+      "step": 1307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030810284493861006,
+      "loss": 1.6757,
+      "step": 1308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030796354916245425,
+      "loss": 1.5378,
+      "step": 1309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030782417943863317,
+      "loss": 1.6727,
+      "step": 1310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030768473586260573,
+      "loss": 1.6966,
+      "step": 1311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003075452185298812,
+      "loss": 1.6096,
+      "step": 1312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003074056275360194,
+      "loss": 1.8107,
+      "step": 1313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030726596297663076,
+      "loss": 1.5413,
+      "step": 1314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030712622494737584,
+      "loss": 1.6315,
+      "step": 1315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030698641354396576,
+      "loss": 1.5972,
+      "step": 1316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030684652886216174,
+      "loss": 1.5292,
+      "step": 1317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030670657099777517,
+      "loss": 1.6484,
+      "step": 1318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003065665400466678,
+      "loss": 1.624,
+      "step": 1319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030642643610475116,
+      "loss": 1.4635,
+      "step": 1320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003062862592679869,
+      "loss": 1.5574,
+      "step": 1321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030614600963238666,
+      "loss": 1.6276,
+      "step": 1322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003060056872940118,
+      "loss": 1.5693,
+      "step": 1323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003058652923489736,
+      "loss": 1.5445,
+      "step": 1324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030572482489343297,
+      "loss": 1.5803,
+      "step": 1325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030558428502360066,
+      "loss": 1.6336,
+      "step": 1326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030544367283573664,
+      "loss": 1.5629,
+      "step": 1327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003053029884261509,
+      "loss": 1.5731,
+      "step": 1328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003051622318912025,
+      "loss": 1.6852,
+      "step": 1329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003050214033273002,
+      "loss": 1.5451,
+      "step": 1330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003048805028309019,
+      "loss": 1.5027,
+      "step": 1331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030473953049851485,
+      "loss": 1.7529,
+      "step": 1332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030459848642669536,
+      "loss": 1.581,
+      "step": 1333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003044573707120491,
+      "loss": 1.5565,
+      "step": 1334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003043161834512308,
+      "loss": 1.4594,
+      "step": 1335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003041749247409439,
+      "loss": 1.6142,
+      "step": 1336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0003040335946779411,
+      "loss": 1.693,
+      "step": 1337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030389219335902383,
+      "loss": 1.5725,
+      "step": 1338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030375072088104234,
+      "loss": 1.6308,
+      "step": 1339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030360917734089564,
+      "loss": 1.4691,
+      "step": 1340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030346756283553136,
+      "loss": 1.6952,
+      "step": 1341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030332587746194584,
+      "loss": 1.7193,
+      "step": 1342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030318412131718385,
+      "loss": 1.4834,
+      "step": 1343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00030304229449833864,
+      "loss": 1.6125,
+      "step": 1344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000302900397102552,
+      "loss": 1.5938,
+      "step": 1345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003027584292270138,
+      "loss": 1.6118,
+      "step": 1346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003026163909689626,
+      "loss": 1.4214,
+      "step": 1347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003024742824256848,
+      "loss": 1.6385,
+      "step": 1348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003023321036945149,
+      "loss": 1.5173,
+      "step": 1349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030218985487283584,
+      "loss": 1.5937,
+      "step": 1350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003020475360580782,
+      "loss": 1.7806,
+      "step": 1351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003019051473477208,
+      "loss": 1.5565,
+      "step": 1352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030176268883929015,
+      "loss": 1.6108,
+      "step": 1353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003016201606303605,
+      "loss": 1.6358,
+      "step": 1354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003014775628185541,
+      "loss": 1.6248,
+      "step": 1355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003013348955015407,
+      "loss": 1.6929,
+      "step": 1356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030119215877703767,
+      "loss": 1.4244,
+      "step": 1357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003010493527428098,
+      "loss": 1.5937,
+      "step": 1358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003009064774966696,
+      "loss": 1.6738,
+      "step": 1359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030076353313647687,
+      "loss": 1.4366,
+      "step": 1360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003006205197601387,
+      "loss": 1.6663,
+      "step": 1361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003004774374656095,
+      "loss": 1.4839,
+      "step": 1362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003003342863508909,
+      "loss": 1.622,
+      "step": 1363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00030019106651403164,
+      "loss": 1.6411,
+      "step": 1364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0003000477780531275,
+      "loss": 1.6131,
+      "step": 1365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002999044210663214,
+      "loss": 1.7369,
+      "step": 1366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000299760995651803,
+      "loss": 1.579,
+      "step": 1367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029961750190780895,
+      "loss": 1.6389,
+      "step": 1368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002994739399326227,
+      "loss": 1.6553,
+      "step": 1369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002993303098245743,
+      "loss": 1.5453,
+      "step": 1370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029918661168204074,
+      "loss": 1.5497,
+      "step": 1371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002990428456034454,
+      "loss": 1.5584,
+      "step": 1372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029889901168725813,
+      "loss": 1.6085,
+      "step": 1373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029875511003199545,
+      "loss": 1.4761,
+      "step": 1374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029861114073622015,
+      "loss": 1.6032,
+      "step": 1375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029846710389854137,
+      "loss": 1.5981,
+      "step": 1376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002983229996176145,
+      "loss": 1.6601,
+      "step": 1377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002981788279921412,
+      "loss": 1.7005,
+      "step": 1378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029803458912086914,
+      "loss": 1.6401,
+      "step": 1379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002978902831025921,
+      "loss": 1.5898,
+      "step": 1380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002977459100361498,
+      "loss": 1.612,
+      "step": 1381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029760147002042816,
+      "loss": 1.8926,
+      "step": 1382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002974569631543585,
+      "loss": 1.5778,
+      "step": 1383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002973123895369183,
+      "loss": 1.5375,
+      "step": 1384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971677492671305,
+      "loss": 1.6365,
+      "step": 1385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029702304244406393,
+      "loss": 1.577,
+      "step": 1386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002968782691668329,
+      "loss": 1.7008,
+      "step": 1387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002967334295345972,
+      "loss": 1.6012,
+      "step": 1388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002965885236465621,
+      "loss": 1.6782,
+      "step": 1389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002964435516019782,
+      "loss": 1.4459,
+      "step": 1390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029629851350014146,
+      "loss": 1.7411,
+      "step": 1391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002961534094403931,
+      "loss": 1.7356,
+      "step": 1392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029600823952211964,
+      "loss": 1.6521,
+      "step": 1393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002958630038447523,
+      "loss": 1.5344,
+      "step": 1394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029571770250776786,
+      "loss": 1.5301,
+      "step": 1395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029557233561068765,
+      "loss": 1.68,
+      "step": 1396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029542690325307813,
+      "loss": 1.4368,
+      "step": 1397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002952814055345506,
+      "loss": 1.5486,
+      "step": 1398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029513584255476107,
+      "loss": 1.5303,
+      "step": 1399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002949902144134101,
+      "loss": 1.7193,
+      "step": 1400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029484452121024315,
+      "loss": 1.5619,
+      "step": 1401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002946987630450502,
+      "loss": 1.5695,
+      "step": 1402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029455294001766544,
+      "loss": 1.6316,
+      "step": 1403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002944070522279679,
+      "loss": 1.5912,
+      "step": 1404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002942610997758806,
+      "loss": 1.5533,
+      "step": 1405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002941150827613711,
+      "loss": 1.562,
+      "step": 1406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002939690012844512,
+      "loss": 1.4581,
+      "step": 1407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002938228554451765,
+      "loss": 1.5636,
+      "step": 1408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029367664534364716,
+      "loss": 1.5124,
+      "step": 1409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029353037108000696,
+      "loss": 1.6057,
+      "step": 1410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002933840327544439,
+      "loss": 1.7397,
+      "step": 1411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002932376304671896,
+      "loss": 1.4797,
+      "step": 1412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029309116431851977,
+      "loss": 1.5348,
+      "step": 1413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002929446344087537,
+      "loss": 1.6984,
+      "step": 1414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002927980408382544,
+      "loss": 1.5099,
+      "step": 1415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029265138370742846,
+      "loss": 1.6399,
+      "step": 1416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029250466311672593,
+      "loss": 1.4532,
+      "step": 1417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002923578791666405,
+      "loss": 1.6414,
+      "step": 1418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002922110319577091,
+      "loss": 1.5307,
+      "step": 1419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029206412159051216,
+      "loss": 1.6637,
+      "step": 1420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002919171481656731,
+      "loss": 1.4079,
+      "step": 1421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002917701117838587,
+      "loss": 1.5175,
+      "step": 1422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000291623012545779,
+      "loss": 1.5826,
+      "step": 1423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029147585055218687,
+      "loss": 1.7214,
+      "step": 1424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029132862590387814,
+      "loss": 1.6212,
+      "step": 1425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029118133870169175,
+      "loss": 1.6883,
+      "step": 1426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029103398904650935,
+      "loss": 1.6964,
+      "step": 1427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002908865770392555,
+      "loss": 1.5076,
+      "step": 1428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029073910278089717,
+      "loss": 1.6896,
+      "step": 1429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029059156637244437,
+      "loss": 1.7404,
+      "step": 1430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029044396791494935,
+      "loss": 1.654,
+      "step": 1431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000290296307509507,
+      "loss": 1.5446,
+      "step": 1432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002901485852572546,
+      "loss": 1.529,
+      "step": 1433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029000080125937195,
+      "loss": 1.5286,
+      "step": 1434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002898529556170808,
+      "loss": 1.5049,
+      "step": 1435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002897050484316455,
+      "loss": 1.635,
+      "step": 1436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002895570798043723,
+      "loss": 1.651,
+      "step": 1437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002894090498366095,
+      "loss": 1.4991,
+      "step": 1438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002892609586297478,
+      "loss": 1.6107,
+      "step": 1439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002891128062852194,
+      "loss": 1.5898,
+      "step": 1440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002889645929044986,
+      "loss": 1.5908,
+      "step": 1441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028881631858910145,
+      "loss": 1.7763,
+      "step": 1442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028866798344058563,
+      "loss": 1.5483,
+      "step": 1443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028851958756055077,
+      "loss": 1.7179,
+      "step": 1444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002883711310506378,
+      "loss": 1.4866,
+      "step": 1445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028822261401252933,
+      "loss": 1.6043,
+      "step": 1446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028807403654794943,
+      "loss": 1.6282,
+      "step": 1447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028792539875866353,
+      "loss": 1.6041,
+      "step": 1448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002877767007464783,
+      "loss": 1.4927,
+      "step": 1449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028762794261324184,
+      "loss": 1.8006,
+      "step": 1450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028747912446084326,
+      "loss": 1.5735,
+      "step": 1451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002873302463912128,
+      "loss": 1.8155,
+      "step": 1452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002871813085063219,
+      "loss": 1.6441,
+      "step": 1453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002870323109081827,
+      "loss": 1.5956,
+      "step": 1454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002868832536988484,
+      "loss": 1.6078,
+      "step": 1455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002867341369804132,
+      "loss": 1.5815,
+      "step": 1456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028658496085501176,
+      "loss": 1.5454,
+      "step": 1457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028643572542481953,
+      "loss": 1.6233,
+      "step": 1458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002862864307920526,
+      "loss": 1.6219,
+      "step": 1459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00028613707705896766,
+      "loss": 1.5719,
+      "step": 1460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002859876643278618,
+      "loss": 1.6911,
+      "step": 1461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002858381927010725,
+      "loss": 1.353,
+      "step": 1462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002856886622809777,
+      "loss": 1.7445,
+      "step": 1463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028553907316999547,
+      "loss": 1.6386,
+      "step": 1464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002853894254705842,
+      "loss": 1.6864,
+      "step": 1465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028523971928524233,
+      "loss": 1.6302,
+      "step": 1466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028508995471650836,
+      "loss": 1.8208,
+      "step": 1467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028494013186696084,
+      "loss": 1.5861,
+      "step": 1468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002847902508392182,
+      "loss": 1.5903,
+      "step": 1469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002846403117359387,
+      "loss": 1.5317,
+      "step": 1470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028449031465982037,
+      "loss": 1.3911,
+      "step": 1471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028434025971360105,
+      "loss": 1.7583,
+      "step": 1472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002841901470000581,
+      "loss": 1.5217,
+      "step": 1473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028403997662200845,
+      "loss": 1.5471,
+      "step": 1474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002838897486823086,
+      "loss": 1.5977,
+      "step": 1475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002837394632838544,
+      "loss": 1.4183,
+      "step": 1476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028358912052958116,
+      "loss": 1.5011,
+      "step": 1477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028343872052246337,
+      "loss": 1.7538,
+      "step": 1478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028328826336551476,
+      "loss": 1.6839,
+      "step": 1479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002831377491617883,
+      "loss": 1.5824,
+      "step": 1480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002829871780143758,
+      "loss": 1.6135,
+      "step": 1481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028283655002640844,
+      "loss": 1.579,
+      "step": 1482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028268586530105593,
+      "loss": 1.5095,
+      "step": 1483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000282535123941527,
+      "loss": 1.5685,
+      "step": 1484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002823843260510694,
+      "loss": 1.5204,
+      "step": 1485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002822334717329692,
+      "loss": 1.5992,
+      "step": 1486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002820825610905514,
+      "loss": 1.5639,
+      "step": 1487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002819315942271795,
+      "loss": 1.5365,
+      "step": 1488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002817805712462554,
+      "loss": 1.5937,
+      "step": 1489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002816294922512197,
+      "loss": 1.7917,
+      "step": 1490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028147835734555114,
+      "loss": 1.6199,
+      "step": 1491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002813271666327669,
+      "loss": 1.5762,
+      "step": 1492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028117592021642204,
+      "loss": 1.6284,
+      "step": 1493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002810246182001103,
+      "loss": 1.4866,
+      "step": 1494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002808732606874631,
+      "loss": 1.642,
+      "step": 1495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00028072184778215004,
+      "loss": 1.6069,
+      "step": 1496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002805703795878787,
+      "loss": 1.4589,
+      "step": 1497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002804188562083944,
+      "loss": 1.5317,
+      "step": 1498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002802672777474803,
+      "loss": 1.6346,
+      "step": 1499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002801156443089573,
+      "loss": 1.6194,
+      "step": 1500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027996395599668404,
+      "loss": 1.5903,
+      "step": 1501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002798122129145566,
+      "loss": 1.5144,
+      "step": 1502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002796604151665086,
+      "loss": 1.638,
+      "step": 1503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002795085628565112,
+      "loss": 1.5247,
+      "step": 1504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002793566560885729,
+      "loss": 1.5199,
+      "step": 1505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002792046949667394,
+      "loss": 1.6022,
+      "step": 1506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002790526795950937,
+      "loss": 1.8061,
+      "step": 1507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027890061007775606,
+      "loss": 1.6278,
+      "step": 1508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027874848651888355,
+      "loss": 1.5396,
+      "step": 1509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002785963090226706,
+      "loss": 1.6383,
+      "step": 1510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027844407769334837,
+      "loss": 1.5914,
+      "step": 1511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027829179263518493,
+      "loss": 1.65,
+      "step": 1512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002781394539524851,
+      "loss": 1.467,
+      "step": 1513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002779870617495905,
+      "loss": 1.6693,
+      "step": 1514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002778346161308794,
+      "loss": 1.4613,
+      "step": 1515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002776821172007667,
+      "loss": 1.7304,
+      "step": 1516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027752956506370366,
+      "loss": 1.5541,
+      "step": 1517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002773769598241782,
+      "loss": 1.6559,
+      "step": 1518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027722430158671444,
+      "loss": 1.5326,
+      "step": 1519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002770715904558729,
+      "loss": 1.372,
+      "step": 1520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027691882653625016,
+      "loss": 1.6196,
+      "step": 1521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027676600993247917,
+      "loss": 1.4266,
+      "step": 1522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027661314074922894,
+      "loss": 1.6936,
+      "step": 1523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027646021909120435,
+      "loss": 1.6493,
+      "step": 1524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002763072450631463,
+      "loss": 1.8153,
+      "step": 1525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002761542187698316,
+      "loss": 1.6097,
+      "step": 1526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002760011403160728,
+      "loss": 1.5904,
+      "step": 1527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002758480098067182,
+      "loss": 1.7151,
+      "step": 1528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027569482734665174,
+      "loss": 1.7408,
+      "step": 1529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000275541593040793,
+      "loss": 1.7196,
+      "step": 1530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000275388306994097,
+      "loss": 1.721,
+      "step": 1531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002752349693115541,
+      "loss": 1.5782,
+      "step": 1532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027508158009819034,
+      "loss": 1.4519,
+      "step": 1533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002749281394590668,
+      "loss": 1.61,
+      "step": 1534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027477464749927987,
+      "loss": 1.4476,
+      "step": 1535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000274621104323961,
+      "loss": 1.5957,
+      "step": 1536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027446751003827683,
+      "loss": 1.5382,
+      "step": 1537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002743138647474289,
+      "loss": 1.4716,
+      "step": 1538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00027416016855665395,
+      "loss": 1.5748,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027400642157122317,
+      "loss": 1.6772,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002738526238964428,
+      "loss": 1.507,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027369877563765383,
+      "loss": 1.6087,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027354487690023176,
+      "loss": 1.5503,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027339092778958684,
+      "loss": 1.7127,
+      "step": 1544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027323692841116354,
+      "loss": 1.6009,
+      "step": 1545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000273082878870441,
+      "loss": 1.648,
+      "step": 1546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002729287792729327,
+      "loss": 1.6575,
+      "step": 1547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002727746297241862,
+      "loss": 1.4992,
+      "step": 1548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002726204303297836,
+      "loss": 1.5329,
+      "step": 1549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027246618119534087,
+      "loss": 1.6748,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027231188242650813,
+      "loss": 1.655,
+      "step": 1551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027215753412896955,
+      "loss": 1.6493,
+      "step": 1552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002720031364084432,
+      "loss": 1.6357,
+      "step": 1553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002718486893706809,
+      "loss": 1.6375,
+      "step": 1554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002716941931214684,
+      "loss": 1.5255,
+      "step": 1555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002715396477666252,
+      "loss": 1.5969,
+      "step": 1556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027138505341200406,
+      "loss": 1.5466,
+      "step": 1557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027123041016349173,
+      "loss": 1.6043,
+      "step": 1558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027107571812700823,
+      "loss": 1.6583,
+      "step": 1559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002709209774085071,
+      "loss": 1.5855,
+      "step": 1560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002707661881139752,
+      "loss": 1.4998,
+      "step": 1561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002706113503494326,
+      "loss": 1.5773,
+      "step": 1562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027045646422093256,
+      "loss": 1.6863,
+      "step": 1563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00027030152983456156,
+      "loss": 1.5917,
+      "step": 1564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002701465472964391,
+      "loss": 1.502,
+      "step": 1565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026999151671271767,
+      "loss": 1.6378,
+      "step": 1566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002698364381895825,
+      "loss": 1.6121,
+      "step": 1567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002696813118332519,
+      "loss": 1.5451,
+      "step": 1568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002695261377499768,
+      "loss": 1.5322,
+      "step": 1569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026937091604604095,
+      "loss": 1.5837,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002692156468277606,
+      "loss": 1.621,
+      "step": 1571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026906033020148443,
+      "loss": 1.6575,
+      "step": 1572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002689049662735939,
+      "loss": 1.5257,
+      "step": 1573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002687495551505026,
+      "loss": 1.7193,
+      "step": 1574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00026859409693865663,
+      "loss": 1.6435,
+      "step": 1575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002684385917445342,
+      "loss": 1.7418,
+      "step": 1576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002682830396746457,
+      "loss": 1.6475,
+      "step": 1577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002681274408355339,
+      "loss": 1.5798,
+      "step": 1578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002679717953337731,
+      "loss": 1.6148,
+      "step": 1579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002678161032759701,
+      "loss": 1.6599,
+      "step": 1580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002676603647687632,
+      "loss": 1.3501,
+      "step": 1581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026750457991882266,
+      "loss": 1.6746,
+      "step": 1582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026734874883285053,
+      "loss": 1.7309,
+      "step": 1583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026719287161758044,
+      "loss": 1.7311,
+      "step": 1584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002670369483797776,
+      "loss": 1.5443,
+      "step": 1585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026688097922623893,
+      "loss": 1.6867,
+      "step": 1586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026672496426379254,
+      "loss": 1.6587,
+      "step": 1587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026656890359929805,
+      "loss": 1.441,
+      "step": 1588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026641279733964634,
+      "loss": 1.4815,
+      "step": 1589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002662566455917596,
+      "loss": 1.6149,
+      "step": 1590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026610044846259114,
+      "loss": 1.5729,
+      "step": 1591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002659442060591252,
+      "loss": 1.6241,
+      "step": 1592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026578791848837734,
+      "loss": 1.4792,
+      "step": 1593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026563158585739385,
+      "loss": 1.8221,
+      "step": 1594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002654752082732519,
+      "loss": 1.6555,
+      "step": 1595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002653187858430594,
+      "loss": 1.6838,
+      "step": 1596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002651623186739551,
+      "loss": 1.5765,
+      "step": 1597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002650058068731084,
+      "loss": 1.5523,
+      "step": 1598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000264849250547719,
+      "loss": 1.6508,
+      "step": 1599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002646926498050175,
+      "loss": 1.7745,
+      "step": 1600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026453600475226467,
+      "loss": 1.7415,
+      "step": 1601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002643793154967517,
+      "loss": 1.59,
+      "step": 1602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026422258214579993,
+      "loss": 1.5654,
+      "step": 1603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002640658048067611,
+      "loss": 1.5435,
+      "step": 1604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026390898358701704,
+      "loss": 1.59,
+      "step": 1605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002637521185939795,
+      "loss": 1.6232,
+      "step": 1606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026359520993509035,
+      "loss": 1.5684,
+      "step": 1607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026343825771782127,
+      "loss": 1.3872,
+      "step": 1608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002632812620496738,
+      "loss": 1.531,
+      "step": 1609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026312422303817934,
+      "loss": 1.6953,
+      "step": 1610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026296714079089887,
+      "loss": 1.7234,
+      "step": 1611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000262810015415423,
+      "loss": 1.4666,
+      "step": 1612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002626528470193719,
+      "loss": 1.4896,
+      "step": 1613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002624956357103952,
+      "loss": 1.8093,
+      "step": 1614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002623383815961719,
+      "loss": 1.6931,
+      "step": 1615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002621810847844104,
+      "loss": 1.585,
+      "step": 1616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00026202374538284817,
+      "loss": 1.6679,
+      "step": 1617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026186636349925216,
+      "loss": 1.4948,
+      "step": 1618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000261708939241418,
+      "loss": 1.619,
+      "step": 1619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002615514727171707,
+      "loss": 1.6404,
+      "step": 1620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000261393964034364,
+      "loss": 1.4362,
+      "step": 1621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002612364133008807,
+      "loss": 1.5449,
+      "step": 1622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026107882062463225,
+      "loss": 1.3076,
+      "step": 1623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026092118611355887,
+      "loss": 1.5811,
+      "step": 1624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026076350987562947,
+      "loss": 1.5561,
+      "step": 1625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002606057920188415,
+      "loss": 1.4853,
+      "step": 1626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002604480326512209,
+      "loss": 1.6639,
+      "step": 1627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002602902318808222,
+      "loss": 1.4128,
+      "step": 1628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00026013238981572803,
+      "loss": 1.6599,
+      "step": 1629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025997450656404944,
+      "loss": 1.6062,
+      "step": 1630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025981658223392574,
+      "loss": 1.6728,
+      "step": 1631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002596586169335243,
+      "loss": 1.6177,
+      "step": 1632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025950061077104064,
+      "loss": 1.7149,
+      "step": 1633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025934256385469804,
+      "loss": 1.5334,
+      "step": 1634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025918447629274806,
+      "loss": 1.7066,
+      "step": 1635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002590263481934697,
+      "loss": 1.498,
+      "step": 1636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025886817966517005,
+      "loss": 1.4971,
+      "step": 1637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002587099708161837,
+      "loss": 1.6817,
+      "step": 1638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025855172175487303,
+      "loss": 1.5568,
+      "step": 1639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025839343258962763,
+      "loss": 1.6308,
+      "step": 1640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025823510342886497,
+      "loss": 1.5502,
+      "step": 1641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002580767343810296,
+      "loss": 1.4908,
+      "step": 1642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002579183255545936,
+      "loss": 1.7024,
+      "step": 1643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002577598770580562,
+      "loss": 1.5162,
+      "step": 1644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002576013889999437,
+      "loss": 1.7059,
+      "step": 1645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002574428614888098,
+      "loss": 1.6551,
+      "step": 1646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025728429463323484,
+      "loss": 1.6692,
+      "step": 1647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025712568854182645,
+      "loss": 1.4655,
+      "step": 1648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002569670433232189,
+      "loss": 1.5454,
+      "step": 1649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002568083590860734,
+      "loss": 1.6795,
+      "step": 1650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002566496359390777,
+      "loss": 1.6098,
+      "step": 1651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025649087399094646,
+      "loss": 1.6315,
+      "step": 1652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002563320733504207,
+      "loss": 1.6027,
+      "step": 1653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000256173234126268,
+      "loss": 1.6203,
+      "step": 1654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002560143564272825,
+      "loss": 1.6865,
+      "step": 1655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00025585544036228454,
+      "loss": 1.5802,
+      "step": 1656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002556964860401208,
+      "loss": 1.4745,
+      "step": 1657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025553749356966405,
+      "loss": 1.5728,
+      "step": 1658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002553784630598134,
+      "loss": 1.4487,
+      "step": 1659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025521939461949387,
+      "loss": 1.5949,
+      "step": 1660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025506028835765637,
+      "loss": 1.4604,
+      "step": 1661
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000254901144383278,
+      "loss": 1.5229,
+      "step": 1662
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025474196280536143,
+      "loss": 1.5213,
+      "step": 1663
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002545827437329352,
+      "loss": 1.3546,
+      "step": 1664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002544234872750535,
+      "loss": 1.6626,
+      "step": 1665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025426419354079627,
+      "loss": 1.6668,
+      "step": 1666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002541048626392686,
+      "loss": 1.5568,
+      "step": 1667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025394549467960153,
+      "loss": 1.4723,
+      "step": 1668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002537860897709511,
+      "loss": 1.5312,
+      "step": 1669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025362664802249885,
+      "loss": 1.7142,
+      "step": 1670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002534671695434515,
+      "loss": 1.7922,
+      "step": 1671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002533076544430408,
+      "loss": 1.53,
+      "step": 1672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002531481028305239,
+      "loss": 1.4422,
+      "step": 1673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025298851481518267,
+      "loss": 1.6576,
+      "step": 1674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025282889050632403,
+      "loss": 1.6241,
+      "step": 1675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025266923001327976,
+      "loss": 1.5965,
+      "step": 1676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025250953344540633,
+      "loss": 1.6764,
+      "step": 1677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002523498009120851,
+      "loss": 1.6505,
+      "step": 1678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002521900325227219,
+      "loss": 1.7018,
+      "step": 1679
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025203022838674716,
+      "loss": 1.7698,
+      "step": 1680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025187038861361585,
+      "loss": 1.5424,
+      "step": 1681
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002517105133128073,
+      "loss": 1.5567,
+      "step": 1682
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025155060259382517,
+      "loss": 1.5819,
+      "step": 1683
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025139065656619734,
+      "loss": 1.5281,
+      "step": 1684
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025123067533947604,
+      "loss": 1.5636,
+      "step": 1685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002510706590232374,
+      "loss": 1.7444,
+      "step": 1686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025091060772708173,
+      "loss": 1.6398,
+      "step": 1687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002507505215606333,
+      "loss": 1.6995,
+      "step": 1688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002505904006335401,
+      "loss": 1.4023,
+      "step": 1689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025043024505547407,
+      "loss": 1.7183,
+      "step": 1690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002502700549361309,
+      "loss": 1.7195,
+      "step": 1691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00025010983038522986,
+      "loss": 1.6163,
+      "step": 1692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002499495715125138,
+      "loss": 1.687,
+      "step": 1693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002497892784277492,
+      "loss": 1.5803,
+      "step": 1694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00024962895124072584,
+      "loss": 1.697,
+      "step": 1695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024946859006125694,
+      "loss": 1.6095,
+      "step": 1696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002493081949991788,
+      "loss": 1.7301,
+      "step": 1697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024914776616435133,
+      "loss": 1.6211,
+      "step": 1698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024898730366665724,
+      "loss": 1.5837,
+      "step": 1699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024882680761600235,
+      "loss": 1.6459,
+      "step": 1700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002486662781223156,
+      "loss": 1.6072,
+      "step": 1701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002485057152955486,
+      "loss": 1.8014,
+      "step": 1702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024834511924567607,
+      "loss": 1.6211,
+      "step": 1703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002481844900826952,
+      "loss": 1.6319,
+      "step": 1704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002480238279166262,
+      "loss": 1.7683,
+      "step": 1705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002478631328575116,
+      "loss": 1.6965,
+      "step": 1706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002477024050154165,
+      "loss": 1.6643,
+      "step": 1707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002475416445004285,
+      "loss": 1.5239,
+      "step": 1708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002473808514226577,
+      "loss": 1.6224,
+      "step": 1709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024722002589223625,
+      "loss": 1.7401,
+      "step": 1710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024705916801931883,
+      "loss": 1.4518,
+      "step": 1711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024689827791408195,
+      "loss": 1.6527,
+      "step": 1712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002467373556867245,
+      "loss": 1.6216,
+      "step": 1713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024657640144746704,
+      "loss": 1.5725,
+      "step": 1714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002464154153065524,
+      "loss": 1.6144,
+      "step": 1715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000246254397374245,
+      "loss": 1.6282,
+      "step": 1716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002460933477608312,
+      "loss": 1.5999,
+      "step": 1717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024593226657661904,
+      "loss": 1.547,
+      "step": 1718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002457711539319381,
+      "loss": 1.5498,
+      "step": 1719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002456100099371395,
+      "loss": 1.4785,
+      "step": 1720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024544883470259594,
+      "loss": 1.7352,
+      "step": 1721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024528762833870145,
+      "loss": 1.7357,
+      "step": 1722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002451263909558713,
+      "loss": 1.5262,
+      "step": 1723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024496512266454224,
+      "loss": 1.5849,
+      "step": 1724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024480382357517196,
+      "loss": 1.5754,
+      "step": 1725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002446424937982394,
+      "loss": 1.8073,
+      "step": 1726
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002444811334442443,
+      "loss": 1.5518,
+      "step": 1727
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002443197426237077,
+      "loss": 1.5512,
+      "step": 1728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024415832144717114,
+      "loss": 1.5013,
+      "step": 1729
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002439968700251972,
+      "loss": 1.4438,
+      "step": 1730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024383538846836896,
+      "loss": 1.57,
+      "step": 1731
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002436738768872905,
+      "loss": 1.5665,
+      "step": 1732
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00024351233539258605,
+      "loss": 1.6741,
+      "step": 1733
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002433507640949006,
+      "loss": 1.5487,
+      "step": 1734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002431891631048994,
+      "loss": 1.4817,
+      "step": 1735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024302753253326812,
+      "loss": 1.418,
+      "step": 1736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002428658724907127,
+      "loss": 1.5073,
+      "step": 1737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024270418308795923,
+      "loss": 1.6726,
+      "step": 1738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002425424644357539,
+      "loss": 1.5927,
+      "step": 1739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024238071664486297,
+      "loss": 1.5454,
+      "step": 1740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024221893982607267,
+      "loss": 1.687,
+      "step": 1741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024205713409018902,
+      "loss": 1.6985,
+      "step": 1742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000241895299548038,
+      "loss": 1.5363,
+      "step": 1743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024173343631046513,
+      "loss": 1.608,
+      "step": 1744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002415715444883357,
+      "loss": 1.8188,
+      "step": 1745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024140962419253465,
+      "loss": 1.6473,
+      "step": 1746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024124767553396618,
+      "loss": 1.5638,
+      "step": 1747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002410856986235542,
+      "loss": 1.6095,
+      "step": 1748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024092369357224183,
+      "loss": 1.5733,
+      "step": 1749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024076166049099147,
+      "loss": 1.3559,
+      "step": 1750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024059959949078468,
+      "loss": 1.6353,
+      "step": 1751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024043751068262225,
+      "loss": 1.533,
+      "step": 1752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00024027539417752398,
+      "loss": 1.5335,
+      "step": 1753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002401132500865285,
+      "loss": 1.4568,
+      "step": 1754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002399510785206935,
+      "loss": 1.5689,
+      "step": 1755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023978887959109546,
+      "loss": 1.6723,
+      "step": 1756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023962665340882954,
+      "loss": 1.5459,
+      "step": 1757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023946440008500966,
+      "loss": 1.7471,
+      "step": 1758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023930211973076823,
+      "loss": 1.5413,
+      "step": 1759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023913981245725629,
+      "loss": 1.4954,
+      "step": 1760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002389774783756432,
+      "loss": 1.5438,
+      "step": 1761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023881511759711667,
+      "loss": 1.4055,
+      "step": 1762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023865273023288287,
+      "loss": 1.7164,
+      "step": 1763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000238490316394166,
+      "loss": 1.6785,
+      "step": 1764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023832787619220853,
+      "loss": 1.6515,
+      "step": 1765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023816540973827075,
+      "loss": 1.6134,
+      "step": 1766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023800291714363124,
+      "loss": 1.5449,
+      "step": 1767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023784039851958632,
+      "loss": 1.521,
+      "step": 1768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023767785397745014,
+      "loss": 1.521,
+      "step": 1769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002375152836285546,
+      "loss": 1.6915,
+      "step": 1770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023735268758424938,
+      "loss": 1.5852,
+      "step": 1771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002371900659559016,
+      "loss": 1.5411,
+      "step": 1772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00023702741885489598,
+      "loss": 1.6783,
+      "step": 1773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023686474639263475,
+      "loss": 1.6284,
+      "step": 1774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002367020486805374,
+      "loss": 1.7522,
+      "step": 1775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002365393258300407,
+      "loss": 1.5532,
+      "step": 1776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002363765779525988,
+      "loss": 1.658,
+      "step": 1777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002362138051596829,
+      "loss": 1.7023,
+      "step": 1778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002360510075627812,
+      "loss": 1.6038,
+      "step": 1779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023588818527339896,
+      "loss": 1.7266,
+      "step": 1780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002357253384030583,
+      "loss": 1.4659,
+      "step": 1781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002355624670632983,
+      "loss": 1.5802,
+      "step": 1782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023539957136567465,
+      "loss": 1.6673,
+      "step": 1783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023523665142175987,
+      "loss": 1.6417,
+      "step": 1784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002350737073431429,
+      "loss": 1.6511,
+      "step": 1785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023491073924142942,
+      "loss": 1.5847,
+      "step": 1786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023474774722824133,
+      "loss": 1.5309,
+      "step": 1787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023458473141521719,
+      "loss": 1.6332,
+      "step": 1788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002344216919140116,
+      "loss": 1.5327,
+      "step": 1789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023425862883629553,
+      "loss": 1.4415,
+      "step": 1790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023409554229375607,
+      "loss": 1.7203,
+      "step": 1791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002339324323980964,
+      "loss": 1.4104,
+      "step": 1792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002337692992610356,
+      "loss": 1.7439,
+      "step": 1793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023360614299430878,
+      "loss": 1.532,
+      "step": 1794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023344296370966686,
+      "loss": 1.657,
+      "step": 1795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023327976151887655,
+      "loss": 1.5843,
+      "step": 1796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002331165365337201,
+      "loss": 1.7402,
+      "step": 1797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002329532888659956,
+      "loss": 1.5051,
+      "step": 1798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023279001862751652,
+      "loss": 1.5654,
+      "step": 1799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023262672593011178,
+      "loss": 1.5687,
+      "step": 1800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023246341088562584,
+      "loss": 1.6106,
+      "step": 1801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023230007360591835,
+      "loss": 1.6117,
+      "step": 1802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023213671420286414,
+      "loss": 1.5951,
+      "step": 1803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023197333278835328,
+      "loss": 1.4736,
+      "step": 1804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023180992947429094,
+      "loss": 1.5587,
+      "step": 1805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002316465043725972,
+      "loss": 1.4072,
+      "step": 1806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023148305759520708,
+      "loss": 1.6154,
+      "step": 1807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002313195892540705,
+      "loss": 1.5905,
+      "step": 1808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002311560994611521,
+      "loss": 1.4672,
+      "step": 1809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00023099258832843123,
+      "loss": 1.6173,
+      "step": 1810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002308290559679019,
+      "loss": 1.5939,
+      "step": 1811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023066550249157262,
+      "loss": 1.511,
+      "step": 1812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023050192801146632,
+      "loss": 1.554,
+      "step": 1813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002303383326396204,
+      "loss": 1.6823,
+      "step": 1814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023017471648808653,
+      "loss": 1.6714,
+      "step": 1815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00023001107966893054,
+      "loss": 1.5731,
+      "step": 1816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022984742229423258,
+      "loss": 1.5146,
+      "step": 1817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022968374447608674,
+      "loss": 1.5385,
+      "step": 1818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022952004632660112,
+      "loss": 1.3938,
+      "step": 1819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022952004632660112,
+      "loss": 1.4386,
+      "step": 1820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022935632795789785,
+      "loss": 1.6079,
+      "step": 1821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022919258948211283,
+      "loss": 1.5689,
+      "step": 1822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022902883101139573,
+      "loss": 1.6072,
+      "step": 1823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002288650526579099,
+      "loss": 1.4643,
+      "step": 1824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022870125453383242,
+      "loss": 1.677,
+      "step": 1825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022853743675135376,
+      "loss": 1.5363,
+      "step": 1826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022837359942267797,
+      "loss": 1.6787,
+      "step": 1827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002282097426600224,
+      "loss": 1.7226,
+      "step": 1828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022804586657561773,
+      "loss": 1.5577,
+      "step": 1829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022788197128170796,
+      "loss": 1.6419,
+      "step": 1830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022771805689055015,
+      "loss": 1.4039,
+      "step": 1831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022755412351441447,
+      "loss": 1.5384,
+      "step": 1832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022739017126558416,
+      "loss": 1.5469,
+      "step": 1833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022722620025635525,
+      "loss": 1.5433,
+      "step": 1834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002270622105990367,
+      "loss": 1.58,
+      "step": 1835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022689820240595027,
+      "loss": 1.6223,
+      "step": 1836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022673417578943043,
+      "loss": 1.7288,
+      "step": 1837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022657013086182414,
+      "loss": 1.5495,
+      "step": 1838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022640606773549092,
+      "loss": 1.744,
+      "step": 1839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022624198652280297,
+      "loss": 1.7007,
+      "step": 1840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022607788733614465,
+      "loss": 1.5063,
+      "step": 1841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022591377028791275,
+      "loss": 1.4925,
+      "step": 1842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002257496354905162,
+      "loss": 1.599,
+      "step": 1843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002255854830563761,
+      "loss": 1.6931,
+      "step": 1844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022542131309792577,
+      "loss": 1.5116,
+      "step": 1845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022525712572761037,
+      "loss": 1.5694,
+      "step": 1846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022509292105788706,
+      "loss": 1.5558,
+      "step": 1847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022492869920122486,
+      "loss": 1.564,
+      "step": 1848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022476446027010455,
+      "loss": 1.6279,
+      "step": 1849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00022460020437701847,
+      "loss": 1.7296,
+      "step": 1850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022443593163447086,
+      "loss": 1.6083,
+      "step": 1851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022427164215497718,
+      "loss": 1.6426,
+      "step": 1852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002241073360510646,
+      "loss": 1.6725,
+      "step": 1853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022394301343527166,
+      "loss": 1.5463,
+      "step": 1854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022377867442014804,
+      "loss": 1.644,
+      "step": 1855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002236143191182548,
+      "loss": 1.6266,
+      "step": 1856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022344994764216406,
+      "loss": 1.6824,
+      "step": 1857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002232855601044592,
+      "loss": 1.5898,
+      "step": 1858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022312115661773425,
+      "loss": 1.6571,
+      "step": 1859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022295673729459456,
+      "loss": 1.6367,
+      "step": 1860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022279230224765608,
+      "loss": 1.7015,
+      "step": 1861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002226278515895456,
+      "loss": 1.4783,
+      "step": 1862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022246338543290063,
+      "loss": 1.5272,
+      "step": 1863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022229890389036921,
+      "loss": 1.4977,
+      "step": 1864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022213440707461,
+      "loss": 1.4683,
+      "step": 1865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022196989509829212,
+      "loss": 1.6906,
+      "step": 1866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022180536807409501,
+      "loss": 1.6675,
+      "step": 1867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022164082611470854,
+      "loss": 1.5075,
+      "step": 1868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022147626933283264,
+      "loss": 1.5927,
+      "step": 1869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022131169784117747,
+      "loss": 1.5544,
+      "step": 1870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022114711175246336,
+      "loss": 1.6529,
+      "step": 1871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022098251117942045,
+      "loss": 1.5055,
+      "step": 1872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022081789623478896,
+      "loss": 1.559,
+      "step": 1873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022065326703131892,
+      "loss": 1.5936,
+      "step": 1874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022048862368177007,
+      "loss": 1.5641,
+      "step": 1875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022032396629891187,
+      "loss": 1.6861,
+      "step": 1876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00022015929499552333,
+      "loss": 1.6423,
+      "step": 1877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002199946098843932,
+      "loss": 1.64,
+      "step": 1878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021982991107831948,
+      "loss": 1.5091,
+      "step": 1879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021966519869010958,
+      "loss": 1.4667,
+      "step": 1880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021950047283258024,
+      "loss": 1.6449,
+      "step": 1881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021933573361855748,
+      "loss": 1.6666,
+      "step": 1882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021917098116087643,
+      "loss": 1.6089,
+      "step": 1883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021900621557238123,
+      "loss": 1.6387,
+      "step": 1884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021884143696592511,
+      "loss": 1.6716,
+      "step": 1885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021867664545437013,
+      "loss": 1.5598,
+      "step": 1886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021851184115058726,
+      "loss": 1.5382,
+      "step": 1887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021834702416745626,
+      "loss": 1.6058,
+      "step": 1888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00021818219461786545,
+      "loss": 1.3473,
+      "step": 1889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021801735261471183,
+      "loss": 1.7165,
+      "step": 1890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021785249827090093,
+      "loss": 1.6397,
+      "step": 1891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002176876316993467,
+      "loss": 1.7583,
+      "step": 1892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021752275301297158,
+      "loss": 1.6599,
+      "step": 1893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021735786232470616,
+      "loss": 1.6129,
+      "step": 1894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021719295974748933,
+      "loss": 1.6891,
+      "step": 1895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021702804539426804,
+      "loss": 1.6068,
+      "step": 1896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021686311937799748,
+      "loss": 1.5649,
+      "step": 1897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021669818181164064,
+      "loss": 1.5479,
+      "step": 1898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021653323280816853,
+      "loss": 1.6326,
+      "step": 1899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021636827248055994,
+      "loss": 1.5672,
+      "step": 1900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002162033009418015,
+      "loss": 1.7149,
+      "step": 1901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021603831830488738,
+      "loss": 1.7113,
+      "step": 1902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021587332468281947,
+      "loss": 1.5232,
+      "step": 1903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002157083201886071,
+      "loss": 1.6994,
+      "step": 1904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021554330493526712,
+      "loss": 1.4806,
+      "step": 1905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002153782790358238,
+      "loss": 1.6832,
+      "step": 1906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021521324260330852,
+      "loss": 1.4143,
+      "step": 1907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021504819575076,
+      "loss": 1.6276,
+      "step": 1908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021488313859122413,
+      "loss": 1.5076,
+      "step": 1909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002147180712377537,
+      "loss": 1.5225,
+      "step": 1910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021455299380340867,
+      "loss": 1.514,
+      "step": 1911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002143879064012558,
+      "loss": 1.6415,
+      "step": 1912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021422280914436866,
+      "loss": 1.6178,
+      "step": 1913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021405770214582765,
+      "loss": 1.5234,
+      "step": 1914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021389258551871975,
+      "loss": 1.6361,
+      "step": 1915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021372745937613865,
+      "loss": 1.6212,
+      "step": 1916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021356232383118443,
+      "loss": 1.6064,
+      "step": 1917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021339717899696373,
+      "loss": 1.5562,
+      "step": 1918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021323202498658942,
+      "loss": 1.5651,
+      "step": 1919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002130668619131808,
+      "loss": 1.731,
+      "step": 1920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002129016898898633,
+      "loss": 1.502,
+      "step": 1921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021273650902976841,
+      "loss": 1.4079,
+      "step": 1922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021257131944603388,
+      "loss": 1.7191,
+      "step": 1923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021240612125180314,
+      "loss": 1.7518,
+      "step": 1924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021224091456022585,
+      "loss": 1.7236,
+      "step": 1925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021207569948445722,
+      "loss": 1.7341,
+      "step": 1926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00021191047613765832,
+      "loss": 1.4978,
+      "step": 1927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002117452446329959,
+      "loss": 1.5701,
+      "step": 1928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021158000508364225,
+      "loss": 1.6809,
+      "step": 1929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021141475760277513,
+      "loss": 1.5072,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002112495023035779,
+      "loss": 1.7324,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021108423929923914,
+      "loss": 1.519,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021091896870295264,
+      "loss": 1.5575,
+      "step": 1933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021075369062791759,
+      "loss": 1.6364,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021058840518733813,
+      "loss": 1.7163,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002104231124944235,
+      "loss": 1.5619,
+      "step": 1936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021025781266238798,
+      "loss": 1.4483,
+      "step": 1937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00021009250580445064,
+      "loss": 1.6906,
+      "step": 1938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020992719203383536,
+      "loss": 1.3692,
+      "step": 1939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020976187146377084,
+      "loss": 1.5346,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020959654420749048,
+      "loss": 1.587,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000209431210378232,
+      "loss": 1.4805,
+      "step": 1942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020926587008923795,
+      "loss": 1.7107,
+      "step": 1943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000209100523453755,
+      "loss": 1.5559,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020893517058503437,
+      "loss": 1.6289,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020876981159633154,
+      "loss": 1.5435,
+      "step": 1946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002086044466009061,
+      "loss": 1.7036,
+      "step": 1947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020843907571202182,
+      "loss": 1.6263,
+      "step": 1948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020827369904294642,
+      "loss": 1.6056,
+      "step": 1949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020810831670695166,
+      "loss": 1.5738,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020794292881731315,
+      "loss": 1.6017,
+      "step": 1951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020777753548731034,
+      "loss": 1.6065,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002076121368302264,
+      "loss": 1.4788,
+      "step": 1953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020744673295934805,
+      "loss": 1.4585,
+      "step": 1954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020728132398796566,
+      "loss": 1.6326,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002071159100293731,
+      "loss": 1.5254,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002069504911968677,
+      "loss": 1.5967,
+      "step": 1957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020678506760375004,
+      "loss": 1.6305,
+      "step": 1958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000206619639363324,
+      "loss": 1.6563,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020645420658889664,
+      "loss": 1.7254,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002062887693937781,
+      "loss": 1.6261,
+      "step": 1961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020612332789128157,
+      "loss": 1.6886,
+      "step": 1962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020595788219472327,
+      "loss": 1.6884,
+      "step": 1963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020579243241742213,
+      "loss": 1.6986,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020562697867269997,
+      "loss": 1.5229,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00020546152107388133,
+      "loss": 1.5998,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002052960597342934,
+      "loss": 1.6359,
+      "step": 1967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020513059476726583,
+      "loss": 1.462,
+      "step": 1968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020496512628613093,
+      "loss": 1.5815,
+      "step": 1969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020479965440422325,
+      "loss": 1.4841,
+      "step": 1970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020463417923487982,
+      "loss": 1.666,
+      "step": 1971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020446870089143974,
+      "loss": 1.5831,
+      "step": 1972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020430321948724446,
+      "loss": 1.6392,
+      "step": 1973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002041377351356374,
+      "loss": 1.5278,
+      "step": 1974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020397224794996413,
+      "loss": 1.694,
+      "step": 1975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020380675804357198,
+      "loss": 1.6439,
+      "step": 1976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020364126552981028,
+      "loss": 1.776,
+      "step": 1977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020347577052203019,
+      "loss": 1.6475,
+      "step": 1978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020331027313358438,
+      "loss": 1.5769,
+      "step": 1979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020314477347782735,
+      "loss": 1.6306,
+      "step": 1980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000202979271668115,
+      "loss": 1.5303,
+      "step": 1981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020281376781780484,
+      "loss": 1.6994,
+      "step": 1982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020264826204025575,
+      "loss": 1.5276,
+      "step": 1983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020248275444882788,
+      "loss": 1.4642,
+      "step": 1984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002023172451568825,
+      "loss": 1.5447,
+      "step": 1985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020215173427778235,
+      "loss": 1.5562,
+      "step": 1986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020198622192489095,
+      "loss": 1.612,
+      "step": 1987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020182070821157312,
+      "loss": 1.5208,
+      "step": 1988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020165519325119436,
+      "loss": 1.6424,
+      "step": 1989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020148967715712113,
+      "loss": 1.4241,
+      "step": 1990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020132416004272065,
+      "loss": 1.4109,
+      "step": 1991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020115864202136093,
+      "loss": 1.626,
+      "step": 1992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002009931232064105,
+      "loss": 1.5725,
+      "step": 1993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020082760371123844,
+      "loss": 1.5214,
+      "step": 1994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020066208364921426,
+      "loss": 1.6411,
+      "step": 1995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000200496563133708,
+      "loss": 1.5666,
+      "step": 1996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020033104227808982,
+      "loss": 1.8666,
+      "step": 1997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002001655211957303,
+      "loss": 1.4865,
+      "step": 1998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002,
+      "loss": 1.5759,
+      "step": 1999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019983447880426972,
+      "loss": 1.6067,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3896,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 7.066888272839967e+17,
+  "trial_name": null,
+  "trial_params": null
+}